N U C L E A T I O N AND ATMOSPHERIC AEROSOLS 1996
R e l a t e d Titles
Books
BRADEN and KOLSTAD Measuring the Demand for Environmental Quality BROWN Air Filtration HOPKE Receptor Modelling for Air Quality Management MASUADA and TAKAHASHI Aerosols, Proceedings of the Third International Aerosol Conference VINCENT Aerosol Science for Industrial Hygienists WILLIAMS and LOYALKA Aerosol Science" Theory and Practice ZANNETTI et al. Air Pollution
Journals
Journal of Aerosol Science Aerosol Science and Technology Annals of Occupational Hygiene Applied Occupational and Environmental Hygiene Atmospheric Environment Atmospheric Research Environmental Health and Pollution Control Journal of Atmospheric and Terrestrial Physics Occupational Health and Industrial Medicine Full details of all Elsevier Science publications and free specimen copies of any Elsevier Science journals are available on request from your nearest Elsevier office.
N U C L E A T I O N AND ATMOSPHERIC AEROSOLS 1996 Edited by
Markku Kulmala Department of Physics University of Helsinki Finland
and
Paul E. Wagner Institut ffir Experimentalphysik Universitfit Wien Austria
Proceedings of The Fourteenth International Conference on Nucleation and Atmospheric Aerosols
PERGAMON
U.K.
Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, U.K.
U.S.A.
Elsevier Science Inc., 660 White Plains Road, Tarrytown, New York 10591-5153, U.S.A.
JAPAN
Elsevier Science Japan, Tsunashima Building Annex, 3-20-12 Yushima, Bunkyo-ku, Tokyo 113, Japan
Copyright 9 1996 Elsevier Science Limited All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means. electronic, electrostatic, magnetic tape, mechanical, photocopying, recording or otherwise, without permission in writing from the publisher.
First edition 1996
Library of Congress Cataloging in Publication Data A catalog record for this book is available from the Library of Congress
British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN 008 0420303
Printed and bound in Great Britain by Redwood Books
The Fourteenth International Conference on Nucleation and Atmospheric Aerosols Helsinki
26- 30 August 1996
International Advisory Committee D. Boulaud, France P. J. Crutzen, Germany N. Fukuta, USA .l. Gras, Australia J. L. Katz, USA M. Kulmala, Finland W. R. Leaitch, Canada
A. A. Lushnikov, Russia E. Meszaros, Hungary T. C. O'Connor, Ireland K. Okuyama, Japan J. H. Seinfeld, USA R. Strey, Germany P. E. Wagner, Austria
Local Organising Committee H. Hirvonen K. [tcimeri M. Kulmala R. Mattsson J.M. Makela A. Laaksonen T. Linnakylci 7". Vesala E Viisanen
Sponsors International Association of Meteorology and Atmospheric Sciences (IAMAS) International Commission on Clouds and Precipitation (ICCP) Committee on Nucleation and Atmospheric Aerosols (CNAA) International Global Atmospheric Chemistry Project (IGAC) University of Helsinki, Department of Physics Finnish Meteorological Institute Finnish Association for Aerosol Research (FAAR) University of Helsinki, Lahti Research and Training Centre Ministry of Education of Finland The Academy of Finland Finnair Maj and Tor Nessling Foundation Neste Foundation
This Page Intentionally Left Blank
CONTENTS PREFACE
...............................................................................................................................................
xxv
Nucleation KATZ J.L., FISK J.A., RUDEK M.M ......................................................................................................... Nucleation of single component supersaturated vapors
1
STREY R., VIISANEN Y ............................................................................................................................ Ternary nucleation of water- n-nonane - n-butanol: Does the amphiphile work as surfactant in vapor phase nucleation?
11
OXTOBY D.W., TALANQUERV., LAAKSONENA ..................................................................................... 21 Density functional theory for binary nucleation HALE B.N., KATHMANN S.M .................................................................................................................. Monte Carlo simulations of small H2SO4-H20 clusters
30
KUSAKA I., WANG Z-G., SEINFELDJ.H ................................................................................................... 34 Monte Carlo simulation of homogeneous binary nucleation: Toward a theory of sulfuric acid-water system ITOH M., TOHNO S .................................................................................................................................. Ion nucleation and growth of sulfuric acid-water aerosol particles: Application of general dynamic equation
38
OKUYAMA K., ADACHI M., KIM T .......................................................................................................... 42 Experimental evaluation of ion-induced nucleation in nanometer-sized particle formation from SOJ H20/N 2 mixtures by a-ray radiolysis KANE D., EL-SHALL M.S ........................................................................................................................ A new technique for ion nucleation using resonance ionization within supersaturated vapors
46
HE F., HOPKE P.K ................................................................................................................................... Experimental study of ion-induced nucleation of volatile organic compounds by radon decay
50
MJkKELA J.M., RIIHELJkM., UKKONEN A., JOKINEN V., KESKINENJ ...................................................... 54 Cluster ion mobility spectra of alcohols SMOLIK J., WAGNER P.E ......................................................................................................................... Joint experiments on homogeneous nucleation. Measurements of nucleation rates in supersaturated n-pentanol vapor
58
SMOLfK J., ZDfMAL V .............................................................................................................................. Homogeneous nucleation rate measurements in 1-pentanol vapour with helium as a buffer gas
61
vii
RUDEK M.M., FISK J.A., KATZ J.L., BEIERSDORFD., UCHTMANN H ..................................................... 65 The homogeneous nucleation of cesium vapor HEIST R.H., BERTELSMANNA ................................................................................................................. Recent experiments concerning the role of non-condensable background gases on nucleation
69
LUIJTEN C.C.M., BOSSCHAART K.J., MUITJENS M.J.E.H., VAN DONGEN M.E.H ................................... 73 Pressure dependence of nucleation rates in binary systems REISS H., KEGEL W.K ............................................................................................................................. Isomorphisms between nucleation theory and microemulsion theory
77
MCCLURG R.B., FLAGANR.C ................................................................................................................. 81 Statistical-mechanical basis for the 1/S correction to classical homogeneous nucleation theory MCGRAW R., LAAKSONENA .................................................................................................................. Scaling properties of the critical nucleus in classical and density functional nucleation theories
85
LAAKSONEN A., MCGRAW R .................................................................................................................. Thermodynamics and phenomenological nucleation theories
89
FORD I.J ................................................................................................................................................... A virial/Fisher model of cluster populations at the critical point
93
SCHMITT J.L., DOSTER G.J., BERTRAND G.L .......................................................................................... 97 The binary nucleation of n-octane and iso-octane PETERS F., RODEMANN T ........................................................................................................................ Homogeneous nucleation rates in binary vapor system Water-n-butanol from pex-tube experiments
101
WYSLOUZIL B.E., WILEMSKIG ............................................................................................................... Transient behavior and time lags in binary nucleation
105
WILEMSKI G., WYSLOUZILB.E ............................................................................................................... Thermocynamic and kinetic consistency of calculated binary nucleation rates
109
ANISIMOV M.P ........................................................................................................................................ Vapor nucleation rate surface topology of the soluble and partially soluble binary mixtures
113
KELKAR M., RAO N.P., GIRSHICKS.L .................................................................................................... Homogeneous nucleation of silicon: Effects of the properties and kinetics of small structured clusters
117
CLEMENT C.F., FORD I.J ......................................................................................................................... Binary contribution to nucleation rates
121
viii
HRUB'~ J., STREY R ................................................................................................................................. 125 Nucleation in ethanol vapor: The effect of dimer formation VOGELSBERGER W .................................................................................................................................. 129 A model for the interpretation of cluster formation in chemical reactions SHNEIDMAN V. A .................................................................................................................................... 133 Time-dependent nucleation in very rapid non-isothermal processes ITKIN A.L ................................................................................................................................................ 137 New microscopic approach to condensation problems PAL P ....................................................................................................................................................... 141 Determination of free energy and rate of nucleation of N204 clusters (Nucleation of rocket fuel in space) KLINGO V.V ............................................................................................................................................ 145 Heterogeneous embryo radius on spherical substrate GOPALAKRISHNAN N., DHANASEKARAN R ............................................................................................. 149 Nucleation mechanism in vapour phase epitaxial growth of binary, ternary and quaternary semiconductors GORBUNOV B., HAMILTON R .................................................................................................................. 153 Multicomponent nucleation on aerosol particles-containing both soluble and insoluble substances SUZUKI K .................................................................................................................................'............... 157 Monte Carlo simulation on the water microcluster in the detailed balance FISENKO S.P ............................................................................................................................................ 161 Nucleation kinetics in gaseous system with big gradients of thermodynamic variables FISENKO S.P ............................................................................................................................................ 165 Photonucleation kinetics FISENKO S.P ............................................................................................................................................ 169 Energy transfer and fluctuations of nucleation rate ITOH M., AONO S., TAKANO H ................................................................................................................ 172 Structure of a metallic microcluster of single and binary-compounds MIKHEEV V. B., LAULAINEN N. S ........................................................................................................... 176 Binary photoinduced nucleation: Transition from the molecular condensation nuclei mechanism to the kinetically controlled binary collision stage (and then to the thermodynamically described behaviour) CHENG R. J .............................................................................................................................................. 180 Nucleation of marine aerosols: A laboratory observation
ix
OLSON T., HAMILL P ............................................................................................................................... 184 A time-dependent solution for the cluster concentrations in homogeneous nucleation KALIKMANOV V.I., VAN DONGEN M.E.H ................................................................................................ 188 Theory of multicomponent nucleation: Effective medium approach DUBTSOV S.N., SKUBNEVSKAYA G.I., SABELFELD K.K., LEVYKIN,A.I.................................................. 192 Aerosol formation induced by W(CO) 6 photolysis in air.Experiment and numerical modelling VOHRA V., HEIST R.H ............................................................................................................................. 195 The flow diffusion nucleation chamber: A quantitative instrument for nucleation research BERTELSMANN A., HEIST R.H ................................................................................................................. 199 How does the wall of the diffusion cloud chamber affect performance ANISIMOV M.P., NASIBULIN A.G ............................................................................................................203 A critical line limitation of embryos Laplass's pressure NOPPEL M ...................................................................... ......................................................................... 208 Nucleation in the presence of air ions and aerosol particles FUKUTA N., WISNIEWSKA M ...................................................................................................................212 Molecular theory of ultramicro clusters and nucleation. I. The surface free energy CLEMENT C.F., HARRISON R.G ...............................................................................................................216 Nucleation from atmospheric fluctuations SOPUCH P., SERVIDA A ...........................................................................................................................220 A new semiphenomenological model for surface tension size dependence LUSHNIKOV A.A., KULMALA M., ARSTILA H ......................................................................................... 225 Nucleation controlled growth of aerosol particles MARS[K F., BLAHA J., LANKAS F ............................................................................................................229 Nucleation and condensation rate measurement by condensation wave SHINAGAWA H., OKUYAMA K .................................................................................................................233 Phase transition of condensate formed by heterogeneous nucleation of condensable vapors onto a cold substrate GRININ A.P., KUNI F.M., KURASOV V.B ................................................................................................237 Kinetic theory of condensation on identical heterogeneous centers KURASOV V.B .........................................................................................................................................241 Condensation on the spectrum of heterogeneous centers with different activities
KOZISEK Z., DEMO P ............................................................................................................................... Modeling of vapour-liquid transient nucleation in binary system
245
BARTELL L.S., KINNEY K.E .................................................................................................................... Motion pictures of nucleation and growth of solid phases in supercooled molecular clusters
249
SINGH B.T ............................................................................................................................................... Computer simulation of prenucleation cluster using percolation theory
250
SINGH B.T ............................................................................................................................................... Biologial application of nucleation theories - I; Nucleation and growth of lung cancer
253
VAN REMOORTERE P., HEATH C., WAGNER P.E., STREY R .................................................................... 256 Effect of supersaturation, temperature and total pressure on the homogeneous nucleation of n-pentanol ZENG X.C., OH K.J ................................................................................................................................. Effect of dimensionality on the temperature dependence of rate of nucleation
260
SCHMELZER J., SLESOV V.V .................................................................................................................... Number of clusters formed in nucleation-growth processes
264
Stratospheric Aerosols and Ice Nucleation CRUTZEN P.J ........................................................................................................................................... The role of particulate matter in ozone photochemistry (stratosphere and troposphere)
268
VALI G .................................................................................................................................................... Ice nucleation - A review
271
PETER TH ................................................................................................................................................ Formation mechanisms of polar stratospheric clouds
280
KARCHER B., LUO B.P ............................................................................................................................ Aerosol production caused by civil air traffic: An overview of near-field interactions
292
SCHUMANN U .......................................................................................................................................... Particle formation in jet aircraft exhausts and contrails for different sulfur containing fuels
296
KINNE S.A., BERGSTROM R.W ................................................................................................................ Latitudinal and temporal variations of the climatic response to enhanced stratospheric aerosol concentrations from the Mt.Pinatubo eruption
300
ZUEV V.V., BURLAKOV V.D., ELNIKOV A.V., MARICHEV V.N .............................................................. 304 Dynamics of development and relaxation of stratospheric aerosol layer after the Mt.Pinatubo eruption based on the observations at Siberian Lidar Station
xi
BARTELL L.S., HUANG J .......................................................................................................................... Kinetics of homogeneous nucleation in large molecular clusters
308
ONASCH T.B., ANTHONY S.E., TISDALER.T., PRENNI A., TOLBERTM.A .............................................. 312 Laboratory studies of sulfate aerosols at low temperature TOLBERT M.A., DISSELKAMPR.S., TIDALER.T., PRENNI A.J., ONASCH T ............................................ 315 Crystallization kinetics of nitric acid dihydrate aerosols: Implications for polar stratospheric clouds KooP T., L u o B.P., BIERMANNU.M., PETER TH .................................................................................... 318 Freezing of binary and ternary solutions of H2SO4,HNO 3 and H20 under stratospheric conditions: Nucleation statistics and experiments KLINGO V.V ............................................................................................................................................ Influence of adsorbed atomic and molecular ions electric field on ice phase formation in clouds
322
MALKINA A.D., PATRIKEEVV.V., KIM N.S., SHKODKINA.V ................................................................ 326 Directed ice nuclei modification by variation of aerosol particles composition You L., YANG S., WANG X., PI J ............................................................................................................ 330 A study on the climate change of ice nuclei concentration in Beijing of 1963 and 1995 KHORGUANI V.G., KHVEDELIDZEZ.V .................................................................................................... 334 On vertical change of concentration of aerosol particles and ice nuclei in atmosphere ADZHIEV A.KH., KALOV R.KH ............................................................................................................... Study of effect of electrical charges and electrical fields on ice-forming activity of aerosols
338
ANTUNA J.-C ........................................................................................................................................... A possible impact of stratospheric aerosols over surface mean temperature trends in Cuba
341
LEVKOV L., BOIN M., KORUBLUEHL., RASCHKEE ................................................................................ 345 A numerical simulation of a contrail HALE B.N., DIMATI'IOD.J ...................................................................................................................... Monte Carlo studies of water/ice adsorbed on model AgI: Effects of lattice shift
349
OHTAKE T ............................................................................................................................................... Ice nucleation on sulfuric acid particles
353
BEARD K.V., MOORE R., OCHS H.T ....................................................................................................... Laboratory studies on evaporation ice nuclei
357
KIKUCHI K., HARADA M., UYEDA H ....................................................................................................... 361 Nucleation characteristics of polycrystalline ice crystals
xii
TIMMRECK C., CRAF H-F., FEICHTERJ., SCHULT I ................................................................................. 365 3D-simulation of the formation and the development of stratospheric aerosol TALPOS S., CUCULEANUV ...................................................................................................................... 369 Vertical diffusion simulation of Be 7 at midlatitudes CHEN Y., KREDENWEISS.M., ROGERS D.C., DEMOTT P.J ..................................................................... 373 Isolating and identifing atmospheric ice-nucleating aerosols: A new technique PLAUDE N.O., POTAPOV E.I., VYCHUZHANINAM.V ............................................................................... 377 Intercomparing results of ice nuclei concentration measurements carried out simultaneously using cloud chamber and filter method TINSLEY B.A ........................................................................................................................................... 381 Does electrostatic charge at cloud tops affect rates of nucleation and sedimentation of ice? PETER TH., MEILINGER S ........................................................................................................................ 385 Size-dependent stratospheric droplet composition in rapid temperature fluctuations. KUZNETSOVA E.M., BOGDAN A., RONTU L ............................................................................................ 389 Numerical simulation of freezing of strong electrolyte solution
Tropospheric Aerosols LUSHNIKOV A.A ...................................................................................................................................... 393 Fractal aggregates in the atmosphere VESALA T ................................................................................................................................................ 403 Phase transitions in Finnish sauna GORBUNOV B., HAMILTONR .................................................................................................................. 407 Water nucleation on aerosol particles containing both soluble and insoluble substances VALKAMA I., POLL,~,NENR ...................................................................................................................... 411 Transport of radioactive materials in convective clouds SPURNY K.R ........................................................................................................................................... 415 On the physical, chemical, and toxic properties of highly dispersed atmospheric aerosols MOLNAR A., OGREN J.A., MI~SZAROSE ................................................................................................. 419 Relationship between light scattering coefficient and chemical composition of atmospheric aerosol particles in Hungary MIRME A., MINKKINEN P., RUUSKANENJ ............................................................................................... 423 Behaviour of urban aerosol, black carbon and gaseous pollutants in urban air: Exploratory principal component analysis
xiii
MACDONALD A.M., LI S.M., ANLAUF K.G., SHARMA S., BANIC C.M., LEAITCHW.R., LIU P.K.S., STRAPP J.W., ISAAC G.A ...................................................................................................... 427 Distribution of methanesulphonate, nss sulphate, SO 2 and dimethylsulphide over the Bay of Fundy and Gulf of Maine SEIDL W., BRUNNEMANN G., KINS L., KOHLERE., REUSSWIG K., RUOSS K., SEILERT., DLUGI R .................................................................................................................................................. Nitrate in the accumulation mode: Data from measurement campaigns in Eastern Germany
431
JAENICNE R., DREILINGV ....................................................................................................................... 435 Time series of the condensation nuclei concentration at the German Neumayer-Station in Antarctica since 1982 VAN DINGENEN R., MANGONI M., PUTAUD J.-P., WATJEN g ................................................................. 439 Ultrafine number size distribution measurements and chemical characterisation of the aerosol over the Atlantic Ocean between 40~ and 40~ PODZIMEK J., CARTER M.A ..................................................................................................................... Particle losses and interaction in Nolan-Pollak counter
443
PRODI F., SANTACHIARa G., MANCINIG ................................................................................................ 447 Measurements of thermophoretic velocities of aerosol particles for incloud scavenging study FASSI FIHRI A., SUHRE K., ROSSET R ...................................................................................................... 451 Internal/external mixing of aerosols by coagulation ALOYAN A.E ........................................................................................................................................... Numerical Modelling of Minor Gas Constituents and aerosols in the atmosphere
455
WILSON J., RAES F .................................................................................................................................. M 3 a multi modal model for aerosol dynamics
458
GE Z., WEXLER A.S., JOHNSTON M.V .................................................................................................... 462 Phase partitioning of aerosol particles during crystallization KOTZICK R., PANNE U., NIESSNER R ...................................................................................................... 465 Heterogeneous condensation properties of ultrafine soot particles (20 nm 5 0 4 - " + H30 +
Studies imply that the above reaction schemes are not the most energetically favorable reaction paths at stratospheric conditions (Chen et al, 1985 and Hofmann et al, 1994). A more probable scenario requires the clustering of multiple water molecules in order to convert the sulfur trioxide molecule into the sulfuric acid molecule and similar clustering of water molecules on each molecule/ion produced in the sequence. The microscopic description of this process is complicated by the mobility of the H § ions and the time scale (-picosecond) of the formation and breaking of hydrogen bonds. '~This work supported in part by the National Science Foundation under Grant No. ATM93-07318.
30
Monte Carlo simulations of small
92S04 -920 clusters
31
MOLECULAR MODEL In this study the rigid water molecules interact via the RSL2 (Rahman and Stillinger 1978) potential and the free H § and (rigid) S O j species interact (mutually and with the H~O) via atom-atom Coulomb plus Lennard-Jones (LJ) potentials. The H20-HzSO 4 interaction energies and the relative size of effective atomic charges in HzSO 4 were determined from quantum chemistry calculations using the GAMESS (1993) software. (See also Kurdi et al, 1989.) The overall scaling of the atomic charges was chosen to give the experimental HzSO4 dipole moment, 2.72 D (Lovas et al, 1981). The LJ parameters, o 0 and e0, for the ijth atomic interaction were estimated from combinatorial rules and adjusted slightly to reproduce realistic H30 § and HSO4 structures and interaction energies consistent with quantum mechanical and thermodynamic estimates ( Mirabel et al, 1991, and Taesler et al, 1969). STATISTICAL MECHANICAL FORMALISM FOR BINARY CLUSTERS The present work employs the Bennett (1976) Metropolis Monte Carlo (Metropolis et al, 1953) free energy calculation technique used previously for small argon LJ cluster (Hale 1982) and RSL2 water clusters (Kemper 1990, Hale 1996). We calculate Helmholtz free energy differences between clusters containing km water and m sulfuric acid molecules and clusters containing k(m-1) water and (m-1) sulfuric acid molecules, where k = ratio of water to sulfuric acid molecules (km/m = k). The Bennett technique allows one to calculate -kT•n[QB/QA], the free energy difference between two systems (called here B and A) with slightly different interaction potentials. QA and QB are configurational partition functions. The Bensemble contains the normal {km, m} cluster with all molecules interacting fully whereas the A-ensemble contains a {k(m-1),(m-1)}cluster with fully interacting molecules plus one free H2SO4 and one free H20. We have simulated small binary clusters at T = 298 K with sulfuric acid mole fraction = 0.5 (k=l). The statistical mechanical formalism assumes that the km water and m sulfuric acid vapor monomers are in equilibrium with the {km, m } binary cluster, k m [ H 2 0 ] + m[H2SO4] {[H~O]km [HzSO4]m}; and the clusters form a non-interacting mixture of ideal "gases", so that the Law of Mass Action is valid:
Qkm,m
[Nkm,m] 1120 Ion
[N1
]
(km)!m ![Q $2~
H2S04 m
[N1
]
Sl2SO4]m
(1)
After some algebra, one obtains the following result for the cluster number distribution: [Nlan,m] : exp[~nm:l[Ckn,, - k
C kn,n =
In
In
11 -
In I2+ k In S, + In S 2 + In ~]]
O]o,t,n Qk(.-l),.-I [Qll]k Q2
(2)
(3)
The ~ , m values are calculated in the Monte Carlo simulations for a series of m values with k fixed and give the free energy differences between the cluster systems A and B. 11 ( 12 ) is the ratio of experimental partial liquid number density to equilibrium partial vapor densities of water (sulfuric acid) above the solution. $1 ($2) is the ratio of the number of ambient vapor monomers to the number of equilibrium vapor monomers above solution for water ( sulfuric acid). Q~,n,m is the {km, m} cluster canonical configurational partition function, and Q11 (Q1z) is the single molecule partition function for water (sulfuric acid). For a fixed k value the simulations are all performed at constant density. Below are shown snapshots of the B and A ensemble for k = 1 and m = 1. ~(k,m) is a combinatorial factor which -.> 1 as m -.> oo
Hale and Kathmann
32
~.;~"
Fig. 1. Snapshot of the B ensemble for k = m = 1 at 298 K.
Fig. 2. Snapshot of the A ensemble for k = m = 1 at 298 K
In order to analyze the configurational free energy differences we use the classical free energy of formation for a binary cluster (see, for example, Wilemski 1988 and Oxtoby 1991) converted to the {km, m} notation:
mFIon,m
=
[36~]m
0
[kin+m] :b'3- km In S 1- m In
S2
(4)
Obulk-liq where o is the binary surface tension. With the assumption of Eq. (4) 5{ln[Nkn~m]} --- - 8m[AFkm,m/kBT] and one obtains the following form for C~m: C~,m 2 [36~]1/3 o k+l = - 3 kB T 2/3
~)bulk-liq
[km+m]_l/3 + k I l + I 2 k+l
(5)
Using Eq. (5) one can extract information about the effective binary surface tension, o, from the slope of C_.~mplotted vs. m"1/3for fixed k. See Fig. 3. In order to calibrate our potential, calculations are performed at 298 K where experimental surface tension and partial vapor pressures are available. RESULTS AND CONCLUSIONS Preliminary results for free energy differences are shown in Fig. 3 for k= 1, for m = 1-6, together with an estimate of C~m from experimental surface tension data (solid line). The calculated C~,.m values for k = 1 are consistent with the rough experimental predictions (extrapolated to small cluster sizes) and show some size effects. A more stringent test of the model potentials depends on C~m for larger cluster sizes which must, in the limit of large m, reproduce the experimental bulk surface tension. As part of the small cluster results, the simulations of the k = m = 1 cluster give an average potential energy of- 17.8 kcal/mole for the HzO-HzSO 4 interaction, compared to -12.8 kcal/mole enthalpy of hydration (Mirabel et al, 1991), and-16.8 kcal/mole from ab initio results ofKurdi et al (1989). The goal of these preliminary studies has been to test the statistical mechanical formalism, to develop a realistic potential model which can be used to study clusters of varying compositions (k values), and to examine some general cluster properties from the Monte Carlo simulations. Root mean square displacements of the atoms indicate that the H + ions are highly mobile and readily bond with both the H20 (to form H3O+) and with the SO4- (to form HSO4 and HzSO 4 ). In this respect the model displays the flexibility essential for modeling the binary system. In progress are calculations for larger m values (with k= 1) and for k = 2, 3, and 4.
33
Monte Carlo simulations of small H2SO4 -920 clusters Binary
H20--H250 T k -
20-
4 Clusters
298 K 1 (0.5 mole
fraction)
4L
J
E
E 10-
(_3
Legend: experimental 1
O.0
1
I
1
d prediction 6 1
0.5 m-1/3
1
1
2 1
m I
1.0
Fig. 3. C~m/[k+l ] vs. m"u3 for k=l and T = 298 K for small binary watersulfuric acid clusters. REFERENCES Bennett, C. H. (1976)J. Computat. Phys. 22, 245. Chen, T. S., and Plummer, P. L. (1985)J. Phys. Chem. 89, 3689. Doyle, G. J. (1961) J. Chem. Phys. 38, 795. GAMESS Quantum Chemistry sot~ware (1993), Ames Laboratory, Iowa State University. Hale, B., and Ward, R. (1982). J. Stat. Phys. 28, 487. Hale, B. (1996)Aust. J. Phys. 49, 425. Hofmann, M., and Von Rague Schleyer, P. (1994). J. Am. Chem. Soc. 116, 4947. Kemper, P. (1990), "A Monte Carlo Simulation of Water Clusters", Ph.D. Thesis, University of Missouri-Rolla Kurdi, L., and Kochanski, E. (1989) Chem. Phys. Lett. 158, 111. Lovas, F. J., Kuczkowski, R. L., and Suenram, R. D. (1981). J. Am. Chem. Soc. 103, 2561. Metropolis, M., Rosenbluth, A., Rosenbluth, M., Teller, A., and Teller, E. (1953). J. Chem. Phys. 21, 1087. Mirabel, P. and Katz, J. L. (1977) J. Chem. Phys. 67, 3763. Mirabel, P., and Ponche, J. L. (1991) Chem. Phys. Lett. 183, 21. Oxtoby, D.W. and Zeng, X. C. (1991) J. Chem. Phys. 95, 5940. Rahman, A., and Stillinger, F. H. (1978)J. Chem. Phys. 68, 666. Reiss, H. (1950) J. Chem. Phys. 18, 840. Taesler, I., and Olovsson, I. (1969). J. Chem. Phys. 51, 4213. Wilemski, G. (1988) J. Chem. Phys. 88, 5124. Yue, G. K., Poole, L. R., Wang, P. H., and Chiou, E. W. (1994). J. Geo. Phys. Res. 99, 3727.
M o n t e C a r l o S i m u l a t i o n of H o m o g e n e o u s B i n a r y N u c l e a t i o n : T o w a r d a T h e o r y of Sulfuric A c i d - W a t e r S y s t e m . Isamu Kusaka, Zhen-Gang Wang, and John H. Seinfeld Department of Chemical Engineering California Institute of Technology Pasadena, CA 91125
ABSTRACT To begin to formulate a molecular theory of binary nucleation of sulfuric a~idwater, Monte Carlo simulation is carried out on the model binary system of dipolar molecules and ion pairs. The reversible work of cluster formation and the cluster size distribution are evaluated for this system.
1. I N T R O D U C T I O N Binary nucleation in the sulfuric acid-water system has received much attention from both experimental and theoretical point of view because of its atmospheric relevance. Yet, the theoretical formulation of the process does not extend beyond the realm of the classical capillarity approximation. One of the key quantity in calculating the nucleation rate is the reversible work of cluster formation. Within the framework of the capillarity approximation, this quantity is given in terms of macroscopic thermodynamic quantities, such as bulk free energy and surface tension. Given questions about the applicability of the classical theory, it is clear that molecular level understanding of the process is indispensable to quantitatively predict the nucleation rate. As a first step toward a molecular theory of binary nucleation of sulfuric acidwater, we performed a Monte Carlo simulation on the model binary system composed of spherical dipolar particles (i.e. H20) and ion pairs. Despite its simplified nature,
34
Monte Carlo simulation of homogeneous binary nucleation
35
the model captures important molecular characteristics of acid-water interaction such as dissociation of sulfuric acid in aqueous solution and the formation of hydrated acid molecules. Thus, the present work is intended to provide a molecular level picture of the process and thereby motivate a development of a molecular theory that renders the quantitative prediction of the nucleation rate feasible.
2. F R E E
ENERGY
SURFACE
OF CLUSTER
THE NUCLEATION
FORMATION
AND
PATH
Under typical atmospheric conditions, the acid concentration is extremely small compared to that of water; hence during the time period required for a cluster to acquire or to lose an acid molecule, the cluster attains partial equilibrium w.r.t, the exchange of water molecules with the surrounding vapor (Shugard and et al. 1974). Unlike the case of homogeneous nucleation in a one component system, the cluster is stable. Therefore, when the supersaturation of water is less than unity, nucleation takes place on the free energy surface of cluster formation along the path that passes trough the traditional saddle point.
3. S I M U L A T I O N
METHOD
In this work, a water molecule is represented by a spherical particle with a permanent dipole # at its center. The sulfuric acid molecule is replaced by an ion pair consisting of two spherical particles each with a point charge +q at its center. In addition to the electrostatic interaction, the intermolecular potential is taken to be Lennard-Jonesian. We used parameter values # = 1.8 Debye and q = lel, e being the charge of an electron. Lennard-Jones parameters are chosen to be suitable for water and sodium chloride (Chandrasekhar and et al. 1984). The system is a spherical region outside of which is assumed to be a continuum vapor with given values of density and dielectric constant appropriate for the pure water
Kusaka et al.
36
vapor. Interaction between the molecules in the system and the surrounding is treated in an approximate manner. Thus, the Lennard-Jonesian interaction produces a repulsive external field in the radial direction very near the system boundary. The electrostatic interaction between the the system and the surrounding is treated by the reaction field method (Barker and Watts, 1973), in which the effect of the surrounding vapor is simply to produce an electric field inside the system in response to the electric field of molecules in the system. To take the full advantage of the partial equilibrium mentioned above, we employ a mixed ensemble in which the number of ion pairs are fixed while that of dipoles is allowed to fluctuate. Therefore, the simulation allows, for a given number of ion pairs, the direct evaluation of the cluster size distribution.
To obtain the full distribution
function, one has to be able to compare the distribution obtained for various number of ion pairs. This requires knowledge of the reversible work of cluster formation along the path of the free energy surface. For the mixed ensemble, this reversible work is given by the increment of the thermodynamic potential
r = U- TS-
#wNw
in forming a cluster from the vapor phase. Here, U, T, S, #w, Nw are respectively, the internal energy, temperature, entropy, chemical potential of the water, and the number of water molecule within the system. In the simulation, r is obtained by integrating U and Nw along a proper thermodynamic path. REFERENCES Barker, J. A. and Watts. R. O., 1973: "Monte Carlo Studies of the Dielectric Properties of Water-like Models," Mol. Phys., 26, 789. Chandrasekhar, J., Spellmeyer, D. C., and Jorgensen, W. L., 1984: "Energy Component Analysis for Dilute Aqueous-Solutions of Li +, Na +, F-, and C1- Ions," J. Am. Chem.
Soc., 106, 903.
Monte Carlo simulation of homogeneous binary nucleation Shugard, W.J., Heist, R. H., and Reiss, H., 1974" "Theory of Vapor Phase Nucleation in Binary Mixture of Water and Sulfuric Acid," J. Chem. Phys, 61, 5298.
37
GROWI~
ION NUCIFATION AND OF SULFURIC ACID-WATER AEROSOL PARTICLES: Application of General Dynamic Equation Masayuki ITOH
Department of Chemical Engineering and Materials Science Faculty of Engineering, Doshisha University,Tanabe, Kyoto 610-03, Japan and Susumu TOHNO Institute of Atomic Energy, Kyoto University, Uji, Kyoto 611, Japan The time evolution of size distribution function was calculated with the general dynamic equation regarding the ion--induced nucleation of sulfuric acid--water system. The nucleation of ion-induced one was still effective in low relative humidity, and even after the initial stage of time evolution, the nucleation was still dominant process as well as the coagulation process. However, the condensation process was not important in thin vapor condition Abstract-
Keyword -ion-induced nucleation, sulfuric acid--water system, general dynamic equation INTRODUCTION Recently the sulfuric acid aerosols in a stratosphere are attracting a public attention. The dynamic behavior of aerosol particles in a stratosphere exerts various influences on global climate. The particles are formed from a photo-oxidation and nucleation process after a decomposition of sulfur in COS and/or volcanic ejecta. The particle formation process in the stratosphere, however, is not only homogeneous heteromolecular nucleation but also ioninduced heteromolecular (heterogeneous) nucleation because of the high ion concentration in the stratosphere. Though many numerical simulations were published on the binary component nucleation of sulfuric acid--water system (SAS) based on Reiss and Doyle [1962] or so on, very few were reported on the ion--induced nucleation of the binary system instead of the practical importance. It is also essential for the prediction of instantaneous variations of size distribution and number concentration of the sulfuric acid--water system in the stratosphere to evaluate global climate change; they are directly relating to the optical property for the transmittance of solar radiation. ION-INDUCED NUCLEATION OF SAS AND THE GENERAL DYNAMIC EQUATION (GDE) The value of AG in ion induced nucleation was estimated from a classical model for the binary system,
AG=
g
nj~.-kZ.ln(
)] +4nr2o +
py
J'=l
f 1,,11, (1
e
r
(1)
ro
where s r, r 0, o, q and (p g/p0) are the chemical potential of j species, the radius of embryo, the radius of ion cluster, tile surface tension of embryo, the number of charge in embryo, and the relative saturation ratio of vapor of species j, respectively. The time evolution of size distribution function after the nucleation was calculated with .J
.
J
.
38
Ion nucleation and growth of sulfuric acid-water aerosol particles
39
simultaneous integro-differential equations (i.e., general dynamic equation) regarding the ioninduced nucleation, condensation and coagulation for a multi-component system. The general dynamic equation applied on our calculation is defined with the following integro-differential equation for a time dependent distribution function f(v,t) and a time dependent coagulation kernel K(v -u,u,t).
V
r
o v.o _ 1 f & Z~o - o-[~(vt)'f(vO ' ]Ov
f K(v. .t u.oau o §
(2)
e (v't)f(v'O] % vAv, O +SNv, t)] +O(v,t) v2f(v,O
where the first and the second terms are the coagulation terms, the third and the fourth terms mean condensation process, the fifth is for the gravitational deposition, the seventh, S is the nucleation term and the final one is the term of particle diffusion. Such stiff equation for temperature gradient and particle number concentration v was solved with the Gear's method. The condensation term was evaluated from the Fuches model. Approximation of size distribution function was carried out by using the J-transformation by Berry [1967]. For the integration on such a stiff system as a general dynamic equation, the Gear's predictor-corrector method [Press, 1992] was applied. RESULTS AND DISCUSSIONS For a saturated vapor pressure of pure sulfuric acid, 3.5xl04mmHg was applied, and the alternative value by Doyle [1962] (10-6mmHg) affects the change of nucleation rate by a factor of three. Both condensation coefficients of sulfuric acid and water for the embryo was set to unity. The dielectric constant e was estimated from
e =78 +23x
(3)
with a mole fraction x of sulfuric acid in the sulfuric acid-water system.
Fig. 1 Perspective projection of free energy surface for cluster formation onto an ion
Itoh and Tohno
40
The temperature in a stratosphere is much lower than the freezing point of water. Because of a lack of exact thermodynamic data on sulfuric acid below the freezing point, however, all the calculations were performed with a temperature higher than the freezing point of water. Fig.1 was a typical example of a perspective projection of free energy surface for cluster formation onto an ion. r0 in Eq.(1) was estimated from the hydrate point. Fig.2 showed computed nucleation rates of sulfuric acid-water system onto an ion cluster of unit charge in various acid vapor activities and relative humidities.
'~I I"
i
ii
i
I ....i"
I' '
~.g1~t ~.'~lo-4f 10-6 ~1 10-1 I' 1
I
10-2 10-3 Sulfuric Acid Activity i
'
10-1
"1 '
i
10-2
10-4 I
10-3
l'
-
Sulfuric Acid Vapor Pressure (Pa.)
Fig. 2 Nucleation rates of sulfuric acid-water system onto an ion cluster of unit charge (T=298K) The nucleation rate of ion--induced one was higher than that of homogeneous one, and was more than unity in high ion concentration even in the case of very low relative humidity. The rate of ion--induced nucleation was proportional to the ion concentration itself and not exceeded the value. However, the possibility of ion--induced nucleation in very low relative humidity is very essential for the nucleation in the stratosphere, because the vapor pressure concerned on the nucleation are generally thin in the zone. Typical examples of size distribution change after the onset of the ion-induced nucleation and that of the homogeneous heteromolecular nucleation were shown in Fig.3. After the initial stage of nucleation, a coagulation process became dominant for both case. In thin vapor system, a condensation was not important, however, ion-induced nucleation was still effective. For thick vapor system where homogeneous nucleation would be set on, breakdown of nucleation happened within a few seconds due to consumption of sulfuric acid vapor by nucleation, and the evolution of size distribution function was separated in the two stage, i.e., coagulation after following condensation. By such mechanisms, the modes of size distribution function by ion-induced nucleation and homogeneous one were bimodal and unimodal, respectively.
Ion nucleation and growth of sulfuric acid-water aerosol particles 107 108
k
0.5 S O.7S
\~_300 ~ 0.4 Q=I llmin "-" V. This means the charged particles are removed at V=300 V before the neutralization by bipolar ions. ~0.2 Besides affecting particle neutralization, the electric " " Q-0 5 l/min field also disturbs the formation of radicals because some 6 0.0 OH radicals are produced by the recombination of H20 + 0.0 0.2 0.4 0.6 0.8 1.0 and e- (M~kePi). This effect can be estimated from voltage SOz c o n c e n t r a t i o n [ p p m ] vs ion-current characteristics (Adachi et al., (1996)). In this case, 30% of OH radical formed by the recombination was Fig. 4 Charged-particle fraction at disturbed. Actual reduction of OH radical by the electric various SO2 concentrations. field was less than 30% because the OH radical formation by the recombination of ions is one of many mechanisms. Therefore, we decided the optimum applied voltage was V=300 V. Figure 4 shows the charged-particle fraction at various SO2 concentrations. The charged-particle fraction decreases with SO2 concentrations. Under no existence of SO2 gas, all particles are charged. This means all particles are generated by ion-induced nucleation at 0 ppm. The reduction of the charged-particle fraction indicates that SO2 molecules react mainly with OH radicals to form H2SO4 ].0
-
9
,
9
!
~
.~_
,
9
.
'
,
'
,
'
p""
O
u
| "O
|
,
|
,
i
,
*
44
Okuyamaet al.
molecules, which then form neutral particles by binary nucleation. The reduction of the charged-particle fraction at flow rate of Q=0.5 I/min is larger than that at Q=l.5 l/min. This difference indicates the particle formation by the ion-induced nucleation is faster than that by the radical formation and binary-nucleation. The total particle number concentration were measured simultaneously with the charged-particle fraction. Figure 5 shows change in total number concentration of particles generated by a-ray radiolysis with the SO2 concentration at flow rates of 0.5, 1.0, and 1.5 l/min. The total particle number concentration is exponentially increases with increasing SO2 concentration, but decreases with flow rate of the mixture.
10 s 10
4
.
~ I/
e-
\
Q=I
a~ ~ 21J/~ ~ 10
~
~810
I/min
Q=1.5 I/min
1"[]/
.
82pCi
lOO
0.0
0.2
0.4
0.6
0.8
1.0
S02 concentration [ppm]
PARTICLE GROWTHFROM IONS
Figure 6 shows the ionization chamber used in the Fig.5 Number concentration of particles particle size distribution measurements. In this generated at various SO 2 concentration. experiment, a 241Am radioactive source of 164 ixCi was used. The size distribution of particles generated in the chamber was measured by the VDMA/FCE system. The electrical mobility distributions of negative and positive ions were measured to establish the measurement system. The average electrical mobilities of positive ions and negative ions agreed with 1.4 cm 2 V -1 s -1 and 1.9 cm 2 V -1 s -1, respectively, which were measured by a mass spectrometer (Bricard). The peak of positive ions was three times larger than that of negative ions. Figure 7 shows effects of water concentration on the electrical ~ in mm mobility distribution of negative ions and charged particles. The electrical mobility distributions showed bimodal peaks due to ions Fig. 6 Ionization chamber for and nanometer-sized particles produced by ion-induced nucleation. particle sizing. When H20 concentration is increased, the peak for ions reduces and the peak for nanometer-sized particles increases. This suggests the H20 molecules acts at initial process of the growth from ions to particles. Although same changes were seen in experiments for positive ions and charged particles, the peak of positively charged particles was smaller than that of negatively charged particles. This difference means the ion-induced nucleation is activated more strongly by negative ions than by positive ions. Figure 8 shows effects of SO2 molecules on the electrical mobility distributions. The SO2 concentration was changed from 0.3 to 3.3 ppm at H20 concentration of 3000 ppm. The electrical mobility distribution changed that the peak for 40 164pci nanometer-sized particles increased with SO2 so 2 =6.7 ppm concentration but the peak for ions decreased Q= 1.5 I/min slightly. This indicates that SO2 molecules works 30 Negative particles at growth process of the nanometer-sized particles. H20=4000 ppm Figure 9 shows the change in the electrical 2000 ) mobility distribution with flow rate of the ~' 20 lOOO SO2/H20/N2 mixture in the ionization chamber. When the flow rate was increased, the peak for ions reduced and the peak for nanometer-sized ]o particles shifted to lower mobility. This means the particle growth depend on the residence time in the ionization chamber. The average diameter of nanometer-sized 10 .2 10-I 10 0 101 particles was obtained from the average electrical mobility of a peak for the particles in Fig.9. Electrical mobility [cm 2 V-is1 ] Figure 10 shows the change in the geometric Fig. 7 Change in electrical mobility distribution mean diameter with the average residence time. with H 20 concentration. . . . . . . . .
0
--
|
. . . . . . . .
,
.......
,
t
. . . . . . .
.
.
.
.
.
45
Ion-induced nucleation in nanometer-sized particle formation The negatively and positively charged particles were generated at two SO2 concentration of 0.3 and 3.3 ppm. The geometric mean diameters for 0.3 and 3.3 ppm increases linearly and nonlinearly, respectively, with the residence time. These linearity and the non-linearity means the particle growth rate at SO2 concentration of 0.3 ppm is constant and that at 3.3 ppm changes with time.
40
. . . . . . .
19
. . . . . . . .
w
. . . . . . . .
|
. . . . . . . .
164pci
0--3000 ppm Q= 1.5 I/min Negative particles
30 ,--,
+Oan}]
=R~
liq
where the superscript (e) refers to the saturated vapor and where
Reiss and Kegel
80 Sm
P vap (e) P vap
(12)
is the supersaturation and R-
Pliq
(e) Pvap
(13)
is the replacement free energy factor. The full quantity in square brackets is the classical expression for Nn. Thus both R and 1/S have their origins in the inclusion of the mixing entropy in the theory, and emerge naturally. The replacement factor prescribed by eq(13) is of the order of 104. It can be shown that if only the mixing entropy associated with the permutation of the drops on the fixed syringes (referring to figure 1) is included in eq (7), the result is
No -exp
/-1"~
[n (~l,liq
~l,vap
/+oa,
where Nd the total number of drops. Since Nn/Nd is the fraction of drops of size n, eq(14) characterizes the polydispersity.. The remaining entropy (associated with the continuum) characterizes the drop population, i.e, the height of the distribution.
REFERENCES 1. Lothe,J.; Pound G.M.J.Chem.Phys. 1962,36,2082. 2. Courtney, W.G.J.Chem.Phys.
1961, 35, 2249.
3. Wilemski, G. J. Chem. Phys. 1995,103,1119: Weakliem, C.L.; Reiss, H. J. Phys. Chem. 1994,98,6408 4. Lothe, J.; Pound G.M.,in Nucleation, ed.Zettlemoyer, A.C.p.112. (Marcel Dekker, New York, 1969) 5. Jouffroy,J.; Levinson, P.; deGennes,P.G.J.Phys (Fr.) 1982,43,1243. 6. Andelman,D.; Cates,M.E., ; Roux,D.Safran,S.A.J.Chem.Phys. 1987, 87 7229. 7. Overbeek, J.T.G.; Verhoeckx, G.J, ; deBruyn,P.L.; Lekkerkerker H.N.W.J.Colloid Interface
Sci. 1987,119,422. 8. Reiss,H. ;Kegel,W.K.; Groenewold, J.; in press, Berichte der Bunsen Gesellschafi. 9. Thomson, W. Phil. Mag. 1871,42,448.
STATISTICAL-MECHANICAL B A S I S F O R T H E 1/S C O R R E C T I O N CLASSICAL HOMOGENEOUS NUCLEATION THEORY
TO
R. B. M c C L U R G and R. C. F L A G A N
Division of Chemistry and Chemical Engineering, MC 210-41 California Institute of Technology Pasadena, CA 91125 A b s t r a c t - There has been considerable confusion as to the origin of the factor 1/S
commonly used to correct the classical theory of nucleation. In the thirty five years since Courtney (1961) first proposed the correction, several authors have proposed alternate justifications. More recently, Weakliem and Reiss (1994)suggested that "one can arrive at an arbitrary factor (including unity)," depending on the details of the model employed. We present a statistical-mechanical justification for the factor 1/S for the limiting case where the saturated vapor can be treated as an ideal gas mixture of n-mers. The derivation suggests the proper resolution of the self-consistency problem raised by Girshick and Chiu (1990). Finally, we show how to use calculated cluster free energies to estimate nucleation rates. We have previously estimated the binding energy and the translational, rotational, vibrational contributions to the free energy for physical clusters of 13 to 147 LennardJones particles. Extrapolations of this data lead to chemical potential, surface tension, and Tolman length estimates for the bulk. Together, the cluster calculations and the bulk property estimates allow one to make unambiguous tests of the capillarity approximation. We find that the capillarity approximation introduces large errors, but inclusion of a size-dependent surface tension (using the Tolman length) nearly reproduces the calculations based on the detailed cluster calculations. Keywords- Nucleation, Correction CLASSICAL NUCLEATION THEORY The rate at which macroscopic droplets spontaneously form in a supersaturated vapor is called the nucleation rate. This process is believed to occur via accommodation to and evaporation of monomers from a growing cluster. According to classical nucleation theory, the flux of clusters (Jn) through a size (n) is
Jn = ~ n~An Cn - En + l Cn + l where an is the accommodation coefficient (commonly assumed to be unity), 13is the flux of monomer through a unit area, An is the surface area of the cluster, Cn is the number concentration, and En is the frequency of monomer evaporation from an n-mer. To determine En+ 1, we follow the approach of Katz (1992). Applying detailed balancing at full thermodynamic equilibrium yields
0 = O~nfleqAnceq -En+lCn+ eq 1 Assuming that the evaporation rate is independent of the saturation ratio leads to the following estimate for En+ 1. En+ ' = a,,fleqAnC, eq / c;~q 1
81
McClurg and Flagan
82 Substituting,
Jn = ~
CnCe q fleqCn+l
(~eq/~)n, and rearranging gives Jn _--- Cn (~e.~_~ln___r~Cn+l([~eqln+l__ an[]AnCeq(fl/fleq) n Ceq Cn+ 1 ~, fl ) For steady state nucleation, J1 = J2 = ... - J. Summing over n yields b 1 -C1 ( - ~ 1 Cb+l([]eqlb+l JZanflAnCnn=l eq (fl/~eq)n - c~q -~bq+l ~ - ~ )
multiplying and dividing by
From the kinetic theory of gases,
[3 = P / (2~'mkT) 1/2 / fleq = p / pvaP =C1/ C~q =S where P is the monomer pressure and S is the saturation ratio. Therefore, determining the nucleation rate (J) reduces to a problem of determining the equilibrium cluster distribution (Ceq). e,o
J=fl/Z[anAnCenqsn]-I n=l This form differs by a factor of 1/S from the original expression developed by Volmer and Weber (1926), Becker and Doring (1935), and Zeldovich (1942). STATISTICAL MECHANICS One rigorous way to determine the equilibrium cluster distribution ( C e q ) is through statistical mechanics. We begin with the partition function for an ideal gas. The partition function for an ideal gas mixture of clusters is e~
Qmix=Hq;n/(in) ! n=l where qn is the partition function for an n-mer and in is the number of n-mers in the mixture. This leads directly to the chemical potential for each component.
l,ln = -kT( C?ln(Qmix) I = -kTln(qn/in ) ~ T,V,iot~ n Thus, at full thermodynamic equilibrium (n/l =/in), the ratio of equilibrium cluster concentrations can be written as C eq _
Cf q
in _ (qn/il) il (ql/il )n
This is the law of mass-action. In its more familiar form, the cluster concentrations are related through a free energy of formation.
The 1/S correction to classical homogeneous nucleation
83
ceq =exp(-AG/kT) AG=
nkTln(q~tlll-kTln(q-~l l=-nl.t+G n
Note that the free energy of formation of a cluster of size one from one monomer is precisely zero. This is the self-consistency requirement as defined by Girshick and Chiu (1990). FREE ENERGY ESTIMATES The most commonly used estimate for AG is the capillarity approximation which is the first term in an asymptotic expansion about the bulk properties. AG-- Act Girshick and Chiu (1990) noted that this approximation does not satisfy the self-consistency condition and proposed subtracting a constant to recover self-consistency. This solution is not in keeping with the asymptotic nature of the capillarity approximation. A better approach is to continue the asymptotic expansion. Tolman (1949) showed that there are additional terms that are important for small clusters.
Rao and McMurry (1990) used this estimate for AG in their nucleation rate calculations. This series maintains the asymptotic character of the original capillarity approximation and suggests that the selfconsistency problem is merely a truncation error. Even better is to calculate cluster free energies using statistical mechanics (McClurg, Flagan, and Goddard, 1995) or Monte Carlo techniques (Weakliem and Reiss, 1994) and then calculate AG directly. CORRECTION FACTORS Depending on the approximation one uses for AG, there are different correction factors to the original nucleation formalism.
J = fJo As noted above, the kinetic derivation of the nucleation rate leads to a factor o f f = 1/S that is common to all of the approaches. The self-consistency correction (Girshick and Chiu, 1990) leads to a factor of exp(A 1cr/kT)/S. We do not recommend this factor because it is not consistent with the asymptotic nature of the capillarity approximation. Rao and McMurry (1990) incorporated the Tolman length into their nucleation rate calculations, but the results are not easily expressed as a correction factor. The various factors discussed by Weakliem and Reiss (1994) are different from 1/S due to several assumptions used in separating the translational degrees of freedom from the internal degrees of freedom. Since these are details concerning the method used in calculating AG, we feel that they should not be incorporated into the nucleation rate formalism. Nucleation rate calculations based on the capillarity approximation, the capillarity approximation with Tolman correction, and direct statisticalmechanical calculation of AG all share the same correction factor.
f= 1/S The methods differ in their estimates of AG. Since these differences can lead to complicated correction factors, they are best expressed as corrections to AG.
McClurg and Flagan
84
We have previously shown that the capillarity approximation introduces large errors into nucleation rate calculations, but the Tolman correction nearly reproduces rates based on direct calculation of cluster free energies. (McClurg, Flagan, and Goddard, 1996) This is despite a (smaller) residual truncation error. ACKNOWLEDGMENTS Partial support of this work was provided by grants from NSF (CHE 94-13930, CTS 91-13191, and ASC 92-17368) and from the International Fine Particle Research Institute.
A C E G J P Qmix S T V eq i int k m n
q r
tr O~
8 (3
g
SYMBOLS cluster area [L2] cluster number concentration [1/L 3] frequency of monomer evaporation from an n-mer [l/t] free energy [mLZ/t2] nucleation rate [ 1/L3t] pressure [m/Lt 2] mixture partition function [-] saturation ratio [-] absolute temperature [T] total volume for mixture [L 3] quantity evaluated at equilibrium conditions (i.e. S = 1) number of n-mers internal contribution (to q) Boltzman's constant monomer mass [m] number of monomers in a cluster [-] cluster partition function [-] cluster radius [L] translational contribution (to q) accommodation coefficient [-] monomer flux [ 1/L2t] Tolman length [L] bulk surface tension [m/t 2] chemical potential [mL2/t 2]
REFERENCES Becker, R. and Doring, W. (1935) Ann. Phys. 24, 719. Courtney, W. G. (1961)J. Chem. Phys. 35, 2249. Girshick, S. L. and Chiu, C-P. (1990) J. Chem. Phys. 93, 1273. Katz, J. L. (1992) Pure Appl. Chem. 64, 1661. McClurg, R. B., Flagan, R. C., and Goddard, W. A. (1995) J. Chem. Phys. 102, 3322. McClurg, R. B., Flagan, R. C., and Goddard, W. A. (1996) Submitted to J. Chem. Phys. Rao, N. P. and McMurry, P. H. (1990) Aerosol Sci. and Tech. 13, 183. Tolman, R. C. (1949) J. Chem. Phys. 17,333. Volmer, M. and Weber, A. (1926) Z Phys. Chem. (Leipzig) 199, 277. Weakliem, C. L. and Reiss, H. (1994) J. Phys. Chem. 98, 6408. Zeldovich, J. (1942)J. Exp. Theor. Phys. 12,525.
SCALING P R O P E R T I E S OF THE CRITICAL NUCLEUS IN CLASSICAL AND DENSITY FUNCTIONAL NUCLEATION T H E O R I E S Robert
McGraw
Environmental Chemistry Division Department of Applied Science Brookhaven National Laboratory, Upton, NY 11973 Ari
Laaksonen
Department of Physics, P.O. Box 9 00014 University of Helsinki Helsinki, Finland Abstract- Scaling relations are developed for the number of molecules in the critical nucleus, g*, and the nucleation barrier height, W* Density functional (DF) calculations for vapor-liquid nucleation confirm these relations and show systematic departure of the ratio W*/(g*Al.t) from its classical value, 1/2, with increasing difference in chemical potential between the supersaturated vapor and bulk condensed phase, A~. Discrepancies between classical and DF nucleation theories and between the classical theory and experiment are interpreted using these results. K e y w o r d s - Nucleation, Clusters, Phase transitions, Densityfunctional theory INTRODUCTION It has been recognized in recent years that homogeneous vapor-liquid nucleation plays a central role in many atmospheric processes. Accordingly, considerable attention has been paid to formulation of phenomenological theories, which try to predict nucleation rates quantitatively starting from macroscopic, measurable properties of fluids (for a review, see [1]). However, at present it is not clear that any of these theories is overall more successful than the classical nucleation theory [2]. On the other hand, more fundamental theoretical approaches that aim to describe the properties of nucleating clusters on a molecular level have been developed [3,4] in parallel with more sophisticated experimental techniques [1]. In this paper, we derive scaling properties for the critical nucleus which are in harmony with novel findings from both experimental and theoretical studies, and which it is hoped will guide the phenomenological efforts in a more productive direction.
85
86
McGraw and Laaksonen
THEORY Consider the nondimensional ratio
W
/ ( g A/l) where
W is the
nucleation barrier height, g is the number of molecules in the critical nucleus, and A~ is the free energy difference between the vapor, at a given saturation ratio S, and the bulk condensed phase driving the phase change. This ratio can be written in the following general form:
W ,
g AFt
where
f
1 =
2
f(g
9 ,A~)
gives the departure from classical nucleation theory.
(1)
In
Refs. 4 and 5 it is shown that the derivative of W* with respect to A~t equals
-g.
Differentiation of Eq. 1 using this result gives: d 9 9 d 9 AFt~--~tg +3g = 2 d A l . t ( f g A/t).
In the classical theory, f =0, and the solution of Eq. 3 for g
(2)
is a
homogeneous function of the form g * = C(T)(A/.t)-3
(4)
in agreement with the Kelvin relation. An interesting result of the DF calculations is that Eq. 4 seems to have a validity beyond the capillarity drop approximation of CNT. This is shown in Fig. 1. To the extent that Eq. 4 is valid, the right hand side of Eq. 2 must also equal zero. For this homogeneity class of systems we have shown that the barrier height differs from the result of classical nucleation theory (CNT) by a constant value at fixed temperature [6]:
9 where density.
W CN T - W
= D ( T ) = n s AIu / 2
(5)
ns is the area of the surface of tension times the superficial A second consequence of homogeneity (Eq. 4) is the following"
87
Scaling properties of the critical nucleus W g Aji
1 2
D(T) C(T)
)2 (6 )
showing recovery of CNT along the coexistence curve, A/z =0, and quadratic dependence in the departure from CNT with increasing A/l. This behavior is seen in the results of the DF calculations shown in Fig. 2. 600
/
500
k
0.
/ ~
400
/
/ /
~CD 3 0 0 /
/
/
~
T=0.8J
k
i 3
I 4
200
100
0 0
1
I 2
500 -
1.0~:/k
400
/1
/"/
300
~ /
/
5
/
/..
T=l.lJk
/ /
200 100 0 0
I 20
I 40
I 60
[i n (f/fo) ]- 3
I 80
100
Fig. 1 Density functional results for g* (markers) and comparison with scaling (lines)from Eq. 4. Temperatures are given in reduced units k T / e where e is the characteristic energy for the Lennard-Jones system of Ref 3. Note for the abcissa that the fugacity ratio is ln(f / f0)= A/l / kT. SUMMARY Measurements of nucleation rate provide a direct probe of W , g , and, for multicomponent systems, nucleus composition [4,5]. In addition to providing an excellent description of the DF results, the scaling theorems (Eqs. 4-6) are shown to be supported by such measurements [6]. Finally, the scaling theorems constrain the departure from CNT and therefore can guide the construction of phenomenological nucleation theories.
88
McGraw and Laaksonen 0.5 :zk _ 9.6) where ridge crossing is expected.
sures makes it possible to have large values of r while the component activities remain roughly comparable. This places the saddle point at a reasonable distance from either pure component axis and permits differences in the growth and steepest descent direction to arise as illustrated in Fig. 3(a). In Fig. 2, we also see that increasing the A-to-B impingement rate ratio by reducing aB at constant aA (= 2.5), leads to a convergence of the two rate predictions because the saddle point moves ever closer to the A axis leaving essentially only one direction through the saddle. Thus, the size of r alone is insufficient to generate a significant difference between the Reiss and Stauffer theories. The location of the saddle point on the free energy surface plays an important role as well, and this is strongly affected by the relative sizes of the equilibrium vapor pressures and other properties of the mixture. At activities in Fig. 2 for which the analytical theories fail, the major flux avoids the saddle point and climbs over a low ridge, as in Fig. 3(b). Ridge crossing has theoretically been ascribed to coupling between the shape of the free energy surface and a large disparity in the impingement rates (r > > 1 or r < < 1) (Trinkaus, 1983). Our result is evidence that the "dynamic" behavior of the free energy surface is also important for producing a situation favoring ridge crossing nucleation as r is varied. The picture of a free energy surface that is fairly insensitive to changes in r may be appropriate for condensed phase nucleation (Greer at al., 1990), but it is not realistic for gas phase nucleation. For gas-liquid systems, the saddle point location can be very sensitive to changes in the vapor composition and, hence, in r. ACKNOWLEDGMENTS This work was supported by the National Science Foundation under Grant No. CHE9502604 (BEW) and the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Geosciences and Engineering (GW). Part of this work was performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under Contract No. W-7405-ENG-48.
112
Wilemski and Wyslouzil 60 .....
,
.,
positive deviator 2, aA = 2.25 aB = 13 , , ~. i , 7, -.,....
,
,-.
60
, ,.,
positive deviator 2, a A = 2.25 aB = 14 , -. , ,
,
55 !i!:, ""-. 5o '::,',,
55 50
"'""-' "" - - .
..... :'" "" " - .
'-'"" "" - . . .
45 40 35
g" 35
-.'-22222-..
"-." ....
"'" ....
30 25
z~ 2
'.
20 ~
:
~
5
~
20
15
15
10
10
5 0 0
10
20
30
40
50
60
70
80
component A molecules per cluster
0
10
20
30
40
50
60
70
80
component A molecules per cluster
Figure 3. Steady state nucleation fluxes superimposed on a contour plot of the free energy surface for the P D2 system at the indicated vapor activities. The saddle point is at the .. The light contours are spaced at 0.5kT intervals relative to the saddle point. The heavy contours start at +2.5kT relative to the saddle and are then spaced at 5kT intervals. (a-left) W / ( k T ) = 47.7 at the analytical saddle point. The major flux flows through the saddle region in a direction distinctly different from the path of steepest descent. (b-right) W / ( k T ) = 45.5 at the analytical saddle point. The major flux bypasses the saddle and climbs over a low ridge that extends from the saddle point toward the lower right corner. R E F E R E N C E S
Greer, A.L., Evans, P.V., Hamerton, R.G., Shangguan, D.K., and Kelton, K.F. (1990) Numerical modeling of crystal nucleation in glasses, J. Cryst. Growth 99, 38. Ko~i~ek, Z. and Demo, P. (1993) Transient kinetics of binary nucleation, J. Cryst. Growth 132, 491. Ko~i~ek, Z. and Demo, P. (1995) Transient nucleation in binary ideal solution, J. Chem. Phys. 102, 7595. Kulmala, M., Laaksonen, A., and Girshick, S.L. (1992) The self-consistency correction to homogeneous nucleation: Extension to binary systems, J. Aerosol Sci. 23, 309. McGraw, R. (1995) Two-dimensional kinetics of binary nucleation in sulfuric acid-water mixtures, J. Chem. Phys. 102, 2098. Reiss, H. (1950) The kinetics of phase transitions in binary systems, J. Chem. Phys. 18, 840. Stauffer, D. (1976) Kinetic theory of two-component ("heteromolecular") nucleation and condensation, J. Aerosol Sci. 7, 319. Trinkaus, H. (1983) Theory of the nucleation of multicomponent precipitates, Phys. Rev. B 27, 7372. Vehkam~ki, H., Paatero, P., Kulmala, M., and Laaksonen, A. (1994) Binary nucleation kinetics: A matrix method, J. Chem. Phys. 101, 9997. Wilemski, G. and Wyslouzil, B.E. (1995) Binary nucleation kinetics. I. Self-consistent size distribution, J. Chem. Phys. 103, 1137. Wyslouzil, B.E. and Wilemski, G. (1995) Binary nucleation kinetics. II. Numerical solution of the birth-death equations, J. Chem. Phys. 10a, 1137. Zeng, X.C. and Oxtoby, D.W. (1991) Binary homogeneous nucleation theory for the gas-liquid transition: A nonclassical approach, J. Chem. Phys. 95, 5940.
VAPOR
NUCLEATION
SOLUBLE
AND
RATE
PARTIALLY
SURFACE
TOPOLOGY
SOLUBLE
BINARY
OF THE
MIXTURES
M i c h a e l P. A N I S I M O V I n s t i t u t e of A t m o s p h e r i c Optics, SB RAS, 1 A c a d e m i c h e s k i i Ave. 634055 Tomsk, Russia. E-mail: A n i s i m o v @ a e r o s o l . k e m e r o v o . s u A b s t r a c t - A t o p o l o g y of n u c l e a t i o n r a t e s u r f a c e s o v e r a P X - d i a g r a m of t h e b i n a r y soluble or p a r t i a l l y soluble solutions a r e discussed. K e y w o r d s - V a p o r nucleation; B i n a r y system; P h a s e d i a g r a m A n idea to s h o w t h e statistical a n d d y n a m i c a l l y p r o p e r t i e s of m o l e c u l a r s y s t e m s as t h e g e o m e t r i c a l figures w a s b o r n m o r e t h a n h u n d r e d y e a r s ago. For e x a m p l e , in Gibbs's t i m e a d i a g r a m of p h a s e s t a t e s w a s n a m e d a g e o m e t r i c a l thermodynamics. P r e l i m i n a r y a t o p o l o g y of a v a p o r n u c l e a t i o n s u r f a c e of b i n a r y v a p o r s w a s d i s c u s s e d b y A n i s i m o v M. a n d A n i s i m o v K. (1994). A t o p o l o g y of n u c l e a t i o n r a t e s u r f a c e s o v e r a P X - d i a g r a m of t h e b i n a r y soluble or p a r t i a l l y soluble solutions a r e d i s c u s s e d in this paper. L e t ' s e x a m i n e a simplest Px - d i a g r a m at a fixed t e m p e r a t u r e T of a b i n a r y solution w i t h an infinite solubility of c o m p o n e n t s . W h e r e P is a total p r e s s u r e a n d x is a c o m p o s i t i o n of a b i n a r y solution. In this case it can be a c i g a r like p h a s e d i a g r a m Fig.l). We will not discuss a vapor depletion and the t i m e lag e f f e c t s of a stationary nucleation rate a f t e r a fast origin of a 1 v a p o r s u p e r s a t u r a t i o n in a Fig. 1 t e s t e d system. W e b e g i n a d r a w i n g of a s u r f a c e of a n u c l e a t i o n r a t e w i t h a r a t e of n u c l e a t i o n of single c o m p o n e n t s A a n d B. In v a p o r e q u i l i b r i u m points (n,m) t h e r a t e s of
113
Anisimov
114
v a p o r n u c l e a t i o n s are e q u a l to zero. Limits of n u c l e a t i o n r a t e s a r e d e f i n e d b y a m e t a s t a b l e r e g i o n b o u n d a r y of t h e single c o m p o n e n t s . T h e points of t h e i r m a x i m u m n u c l e a t i o n r a t e s (c, d) are c o n n e c t e d b y a line (ced) of a m a x i m u m of a b i n a r y n u c l e a t i o n rate. A n u c l e a t i o n of a b i n a r y v a p o r w i t h composition Xo is b e g a n in p o i n t ' f . W h e r e t h e p o i n t f is a first p o i n t of a p r e s s u r e i n t e r v a l (fg) of a h e t e r o g e n e o u s p h a s e e q u i l i b r i u m at t h e composition xo. A c o n d e n s e d p h a s e has a composition x 1 in a p o i n t r. For e a c h point of a line (nfm) t h e r e is a point of a line (nrgm). A line ( n r g m ) of an e q u i l i b r i u m of a liquid a n d t w o p h a s e s states is g e o m e t r i c a l place of points of zero r a t e of n u c l e a t i o n a n d is a b o u n d a r y of a n u c l e a t i o n r a t e surface. O b v i o u s l y a s u r f a c e of n u c l e a t i o n r a t e is d r a w n on a c o n t o u r (cedngm). In a v a p o r labile s t a t e t h e compositions of t h e vapor, a n d c o n d e n s e d p h a s e s h a v e a coincidence. A n a d v a n c e m e n t f r o m a point r p r e s s u r e to an e one gives a line (re) of a v a p o r n u c l e a t i o n r a t e at a t e m p e r a t u r e of n u c l e a t i o n T a n d a v a p o r c o m p o s i t i o n xo. A line (fe) is a p r o j e c t i o n of a line (re) on a state d i a g r a m section at a c o m p o s i t i o n Xo. It does not k n o w n a compositions of critical e m b r y o s in last case. T h e n e x t p r e s e n t a t i o n of a n u c l e a t i o n r a t e s u r f a c e can be p r e s e n t e d if all e x p e r i m e n t a l v a p o r n u c l e a t i o n r a t e s for t h e d i f f e r e n t v a p o r compositions are a t t r i b u t e d to s t a t e d i a g r a m sections at points c o r r e s p o n d i n g to initial v a p o r compositions. In this case a n a u g h t r a t e s points of a v a p o r n u c l e a t i o n b e l o n g a l o w e r line of a c i g a r - l i k e d i a g r a m . W e r e c e i v e a s u r f a c e in a space J X P . T h e s e s u r f a c e s can be r e c o n s t r u c t e d to a space JAi d i s c u s s e d b e f o r e ( A n i s i m o v a n d A n i s i m o v , 1994). W h e r e Ai is an i - t h v a p o r activity.
Fig. 2
N o w it is clear t h e r e are t w o w a y s of a p r e s e n t a t i o n of n u c l e a t i o n r a t e s of a b i n a r y v a p o r nucleation. P r e s e n t a t i o n of n u c l e a t i o n rates with u n k n o w n critical e m b r y o s c o m p o s i t i o n s is usual for e x p e r i m e n t a l results like A n i s i m o v a n d A n i s i m o v (1994). In this case a s u r f a c e of n u c l e a t i o n r a t e is d r a w n on a c o n t o u r (cednfm). A line of
Vapor nucleation rate surface topology an e q u i l i b r i u m of a gas a n d h e t e r o g e n e o u s p h a s e states gives n a u g h t r a t e s of a v a p o r n u c l e a t i o n a n d is a b o u n d a r y of a n u c l e a t i o n r a t e s u r f a c e for u n k n o w n critical e m b r y o s compositions. It m a s t be m a r k e d t h e first a n d s e c o n d w a y s of a n u c l e a t i o n r a t e s u r f a c e p r e s e n t a t i o n h a v e some d e c a d e s of a d i f f e r e n c e of a v a p o r n u c l e a t i o n rates at low n u c l e a t i o n rates. A s t a t e d i a g r a m is like a d r o p if a t e m p e r a t u r e n u c l e a t i o n v a l u e T is b e t w e e n t h e critical t e m p e r a t u r e s of c o m p o n e n t s (Fig.2). A top v a l u e (point c) of a n u c l e a t i o n r a t e of a h i g h critical t e m p e r a t u r e c o m p o n e n t B is c o n n e c t e d b y t h e line of m a x i m u m r a t e s of a b i n a r y v a p o r n u c l e a t i o n w i t h a critical p o i n t at a t e m p e r a t u r e T. A v a p o r r a t e n u c l e a t i o n at a critical point is e q u a l to zero. It is d u e a c o i n c i d e n c e of a critical point w i t h an e q u i l i b r i u m line h a v i n g a zero n u c l e a t i o n r a t e (Anisimov, 1995). F u r t h e r d e s c r i p t i o n is t h e s a m e as for a c i g a r like d i a g r a m . A s u r f a c e of a n u c l e a t i o n r a t e for a b i n a r y s y s t e m w i t h a p a r t i a l solubility of c o m p o n e n t s a n d a p e r e t e c t i c (or eutectic) s t a t e d i a g r a m (Fig.3). F o r solution compositions is a p p l i c a b l e a d e s c r i p t i o n for c i g a r - l i k e d i a g r a m . Let's discuss a n u c l e a t i o n of a v a p o r c o m p o s i t i o n x o in a r e g i o n of h e t e r o g e n e o u s solutions. P r e s s i n g of a v a p o r w i t h c o m p o s i t i o n Xo at a fixed t e m p e r a t u r e leads a new phase with composition x l in point f. T h e low p a r t of a n u c l e a t i o n r a t e c u r v e b e l o n g to a s u r f a c e of solutions till a composition at p o i n t s. B e t w e e n points s a n d v t h e r e a r e a h e t e r o g e n e o u s solutions w i t h d i f f e r e n t ratios of solution c o m p o s i t i o n s x s a n d Xv. A n i n t e r v a l sv of compositions of a v a p o r leads to a t w o c h a n n e l n u c l e a t i o n ( R a y A.K. et al., 1986). A r a t e of n u c l e a t i o n is a s u m of r a t e s of b o u n d points of soluble states. A r e s u l t v a p o r n u c l e a t i o n r a t e is d e t e r m i n e d b y a r e l a t i v e c o n t r i b u t i o n of t w o c h a n n e l s of v a p o r nucleation. A p r o d u c t i v i t y of t h e c h a n n e l s s u b m i t s to t h e l e v e r a g e rule for a v a p o r n u c l e a t i o n rate. A f o r m w h e n all e x p e r i m e n t a l v a p o r n u c l e a t i o n r a t e s are a t t r i b u t e d to points w i t h t h e initial v a p o r c o m p o s i t i o n is closer to a s t a n d a r d of an e x p e r i m e n t results p r e s e n t a t i o n ( S t r e y a n d W a g n e r , 1 9 8 8 ; A n i s i m o v et al., 1987). This s u r f a c e has an a n g l e inflexion at t h e e u t e c t i c point. It is u n u s u a l b e h a v i o r f r o m t h e position v i e w of a classical
115
116
Anisimov nucleation theory. Both forms of nucleation line at h i g h e s t nucleation rates.
rate surfaces coincide w i t h spinodal
For a nucleation t e m p e r a t u r e h i g h e r a critical one of a single c o m p o n e n t of a solution t h e v a p o r compositions at a left side of a critical point are i n t e r e s t i n g for an analysis (Fig.4). A v a p o r u n d e r a pressing don't m e e t s w i t h a p h a s e e q u i l i b r i u m lines in this region of p a r a m e t e r s . In supercritical conditions w e h a v e a h o m o g e n e o u s fluid state. Fluid pressing u n d e r an i s o t h e r m a l condition can lied a nucleation of a h e t e r o g e n e o u s fluid or crystalline particles. H e r e t h e fluid states are a supercritical states. At p r e s s u r e s h i g h e r t h a n an eutectic point t h e single surfaces of liquid nucleation rates restrict a region of nucleation of liquids w i t h a h e t e r o g e n e o u s composition. If a t e m p e r a t u r e of nucleation is b e t w e e n critical t e m p e r a t u r e s of t h e m i x t u r e c o m p o n e n t s , w e can h a v e a nucleation into a supercritical fluid state. A r e g u l a r v a p o r n u c l e a t i o n in a gas m e d i u m is t h e nucleation into a h e t e r o g e n e o u s supercritical fluid state. It seems t h a t a solution d i a g r a m s d e t e r m i n e some principal points of a nucleation theory. A c c o u n t m a s t be t a k e n of solution d i a g r a m s for i n t e r p r e t a t i o n of e x p e r i m e n t a l results on nucleation. A c k n o w l e d g e m e n t s - A u t h o r would like to a c k n o w l e d g e T h e Russian F u n d a m e n t a l R e s e a r c h F u n d for t h e g r a n t 94-03-09947. Russian A c a d e m y of Science is a c k n o w l e d g e d for t h e State Scientific Scholarship. REFERENCES Anisimov, M.P. a n d Anisimov, K.M. (1994). F o u r t h I n t e r n a t i o n a l Aerosol Conference. Abstracts. A A A R , Los Angeles. V1. P.109-110. Anisimov, M.P. (1995)In Aerosols. T h e i r G e n e r a t i o n , B e h a v i o r a n d Application Univ of East Anglia, Norwich. P201-215 Anisimov, M.P., Vershinin, S.N., Akseniov, A.A., Sgonnov, A.M., Semin, G.L. 9 (1987) Kolloidn.Zh.(Russian) V.49, P.842-846 Ray, A.K., M.Chalam, L.K.Peters (1986) J . C h e m . P h y s . v.85 p.2161-2168 S t r e y , R., W a g n e r , P.E.J. (1988) Aerosol Sci. V.19. P. 813-816
To be presented at 14th International Conference on Nucleation and Atmospheric Aerosols, Helsinki, Finland, Aug. 26-30, 1996.
HOMOGENEOUS NUCLEATION OF SILICON: EFFECTS OF THE PROPERTIES AND KINETICS OF SMALL STRUCTURED CLUSTERS
M. KELKAR, N. P. RAO and S. L. GIRSHICK Department of Mechanical Engineering, University of Minnesota, Minneapolis, Minnesota, USA
Abstract- We examine the effects of the properties and kinetics of very small clusters on homogeneous nucleation of a chemically bonded substance such as silicon for which clusters may exhibit solid-like structure. Stepwise changes in cluster free energy are seen to differ significantly from the capillarity model, and this can strongly affect the value of the critical nucleus size. In addition the requirement of third-body stabilization for growth of the smallest clusters can substantially slow the rates of dimer- or trimerization, and can introduce a pressure dependence on nucleation rates under conditions where these processes are rate-limiting. Keywords - Homogeneous nucleation; Silicon; Cluster properties; Three-body kinetics INTRODUCTION The purpose of this paper is to illustrate how the properties and kinetics of very small clusters may affect the dynamics of homogeneous nucleation. As an example we consider silicon, both for its practical interest and because among chemically bonded substances silicon clusters have been relatively well studied. We find that stepwise free energy changes associated with changes in the geometrical configuration of a growing cluster can drastically affect a nucleation calculation, and can make the concept of a "critical nucleus" ambiguous. In addition we consider the requirement of third-body stabilization for growth of the smallest clusters. Under some circumstances this can slow nucleation rates and can introduce a pressure dependence which does not appear in classical theory. NUCLEATION THEORY FOR DISCRETE CLUSTERS Following the usual mathematical formalism of classical theory, one can show that under conditions of a steady-state nucleation current the homogeneous nucleation rate J can be written {c[
(ply-'expl_AGOl]ll-'
(1)
where g denotes the number of monomers comprising a cluster; na is monomer number density; kg is the forward rate constant for the reaction
Ag + A1--->Ag+l" Pl is monomer partial pressure;
p0 is unit pressure
(taken in our calculations as 1 atm); AGO is the free energy change associated with forming a g-mer from
117
118
Kelkar et al.
g vapor phase monomers at unit pressure; kT is Boltzmann's constant times temperature; and the upper index G of the summation is arbitrary but large enough so that subsequent terms count negligibly. Eq. (1) is not the usual expression for J but it is essentially identical. Classical theory does not dwell on J expressed as a summation but instead immediately converts the summation to an integral over a continuous cluster size spectrum. However the summation form is more appropriate for incorporating properties of discrete clusters (Hoare et al., 1980). Further, the use of AG~ seems desirable in that most tabulations of cluster properties are presented at unit pressure. (Note that the surface tension in classical theory is equated with the formation free energy at equilibrium vapor pressure.) Finally, whereas classical theory assumes unity sticking coefficients to evaluate the forward rate constants kg, the summation form allows one to incorporate any available information for size-dependent rate constants. This may be particularly important at the smallest cluster sizes for which sticking coefficients may be far from unity. THERMODYNAMIC PROPERTIES OF SILICON CLUSTERS A more detailed discussion of our method for calculating AG~ for silicon clusters up to size 20 appears in Girshick et al. (1996). Properties of Si2 and Si3 are tabulated in the J A N A F Thermochemical Tables (1985). For larger clustei's we have used information on ground-state structure, binding energy, moments of inertia and vibration frequencies calculated using ab initio quantum mechanics up to size 10 (Raghavachari and Rohlfing, 1988), and molecular dynamics simulations up to size 20 (Chelikowsky et al., 1991). These data were then used to estimate enthalpies and entropies of formation using simplified statistical mechanical expressions (McQuarrie, 1976). Figure 1 shows the stepwise free energy -so change AG'g_~.g 0 at 300 K obtained from these 9 atomistic .
atomistic data compared to the corresponding values as obtained by the classical capillarity model. It is seen that the capillarity model does remarkably well at putting a smooth monotonic curve through the atomistic data, but that the atomistic data deviate sharply about this curve. This effect is primarily due to the dominant contribution to AG of the binding energy, which is affected by variation with cluster size in the cluster's ground-state geometric configuration.
.
.
.
i
.
.
.
.
v
1
9
9
~. -,oo n
9
?
~ -,so
-
9
~
i
~
~
,
,
o
o
,
,
o
o
i
.
capillarity
o
o
o
o
~
1
.
~
o
o
-200
. -250
0
.
.
.
| 5
.
.
.
10
Cluster size,
15
.
.
. 20
g
Figure 1. Formation free energy calculated using atomistic data, comparedto the capillaritymodel.
THIRD-BODY STABILIZATION EFFECTS Monomer-monomer collisions form a dimer with enough excess energy to redissociate if not stabilized by collision of a third body within a characteristic time associated with the vibrational period of the excited dimer. The same holds for condensation of monomers on larger clusters, except that as cluster size increases the ability of the excited complex to absorb this excess energy rapidly increases due to the increasing number of vibrational modes. We have accounted for this phenomenon as follows. Let kg in Eq. (1) denote the effective forward rate constant for the overall reaction Ag + A~ ~ Ag.l, regardless of whether mediation of a third body is required. Theoretical rate constants for silicon dimer formation by three-body recombination were calculated by Martin et al. (1990), with either silicon or argon as the third body. To convert this to an effective two-body rate constant k~, one simply multiplies by the number density of the third body, thus introducing a pressure dependence.
Homogeneous nucleation of silicon
119
For the reaction Si2 + Si ~ Si3, Gai et al. (1988) reported a theoretical calculation of the rate constant k~ for forming the excited Si3 complex, as well as a range of average lifetimes for the excited complex. To convert these data to an effective two-body rate constant k2, let k 2
k;('~lifetime/'~collision) ,
=
where "rlimi,,~ = 5 ps and ~collision is the time between collisions with any stabilizing third body. The resulting values for k~ and k2 are shown in Figure 2. It is seen that the effective rate constant for dimerization is lower than the gas-kinetic collision rate by a factor of 103 at 1 bar total pressure and by 106 at 1 mbar. The effective rate constant for trimerization is lower than the gas-kinetic value by factors ranging from 10-100 at 1 bar to 104 -105 at 1 mbar. Figure 3 suggests an interpolation of these results to the gas-kinetic values expected for larger clusters. Here we have arbitrarily assigned sticking coefficients of 0.5 and 0.9, respectively, to the rate constants k3 and k4 (i.e., for formation of the tetramer and pentamer), while condensation to larger clusters is assumed to have unity sticking coefficient. These values are used in the calculations reported below. 1016
oI
i
t'o
10~4
"~' o
& m
0 e c
.~.
'
i
'
i
'
1
Si 2 + Si = Si a (upper curve), gas kinetic Si + Si = Si 2 (lower curve), gas kinetic .........
10's
% + s , . f i ; s i ~ ; k~:'i'6a/ ................
10 ~2
II (n
."--'
+
'~
10-1S
=-
L.
2 10"
107 200
O
0
O
0
0
O
[]
0
O
0
0
0
O
'o
3
10 'e
=u
o
Si 2 + Si + M = Si a + M, 1 mbar
109
UJ
O
10 17
1 0 10
10 e
0
Si + Si + Ar = Si 2 + Ar. P = I bar
I
E
0 o
O~ a) - S i -+ Si + Ar = Si= + Ar, P = 1 mbar
~
,
I 400
,
t 600
,
l 800
Temperature
~ ,
/ 1 O00
1200
lO-,e
I
o 10-19 0 _ ,
,
,
,
l
5
[K]
Figure 2. Effective 2-body rate constants for 3-body reactions forming Si2 and Si3.
. . . .
J
10
. . . .
Cluster size,
T= 1000K I P = 100 kPa 3rd body = Argon 1
~5
20
g
Figure 3. Size-dependent forward rate constants incorporating 3-body dimer- and trimerization.
EFFECT ON NUCLEATION DYNAMICS From classical theory the summand in Eq. (1) is expected to be dominated by terms located around the critical size. In other words, nucleation is rate-limited by the rate of formation of the critical size cluster. Figures 4 and 5 show values of the summand in Eq. (1) for saturation ratios of 10 and 100, respectively. These calculations were done for a temperature of 1200 K and a total pressure of 1 bar. The estimates of AG~ m a y b e less accurate at 1200 K than at 300 K because of the potential appearance of isomers which may be more stable than the ground state structure assumed in our calculations. However these calculations are adequate for our purpose of illustrating the qualitative effect of cluster structure on nucleation dynamics, and 1200 K is a more interesting temperature than 300 K because of the extremely low equilibrium vapor pressure of silicon, which even at 1200 K is only about 10 -12 atm. For the capillarity model we used the self-consistent form of AG suggested by Girshick and Chiu (1990); this sets AGO to zero (which it must equal by definition, but which classical theory neglects to do) but otherwise has the same stepwise AG~
as in the usual classical version.
Figures 4 and 5 both show values for the summand in Eq. (1) which are drastically different according to whether one uses the atomistic data or the capillarity model. At a saturation ratio S = 10 the atomistic data show a peak at g = 2, whereas classical theory predicts a critical size g * of 118. Referring to Figure 1, it is plausible (if not assured) that for cluster sizes larger than 20 but smaller than 118 the step-
Kelkar et al.
120
30
S=10 (g'c, = 118) 40
capillarity
'7,
E
i,._1 10
t,.-
[]
30
[]
o
[]
O
2o
9
~
lo
ioo
9
15)
1
.
9
9
9
9
9
5
I 10
,
,
i o
o
9 o
9 o
,-~ o
9
9
9
9
0
9
15
2O
Cluster size, g Figure 4. Summandin Eq. (1), comparing atomistic data to the capillarity model, for a saturation ratio of 10.
1
-
'
o
o
o
- ~ - o
o
9
9
9
atomistic 9
9
-10
T = 1200 K, P =
T = 120OI K, P = 1 0 0 k P a , . . . . . . . .
[]
capillarity
v
atomistic 9
. o
9
E
9
. o
o
r
9
.
El
D
o
E
o
9 00
'
(g'c, =
@@
f3
O
[]
r._l o
[]
[]
O
E E
[]
O
s
"200
....
,
,
100kPa 1
5
,
, _,
,
1 10
,
__,__
,
,
1 15
Cluster size, g
Figure 5. Summandin Eq. (1), comparing atomistic data to the capillarity model, for a saturation ratio of 100.
wise free energy changes will converge to the value given by capillarity. In that case the value of the summand for the atomistic data shown in Figure 4 could rise above the value at g = 2, in which case the summand would indeed peak at 118. Note, however, that the value of the summand at this peak, and hence the nucleation rate, might be quite different than predicted by classical theory. It thus appears that the properties of small clusters could affect nucleation rates even when critical sizes are relatively large. For the case of S = 100, shown in Figure 5, the classical theory predicts a critical size of 15, whereas the atomistic results indicate that the nucleation rate is almost completely governed by the rates of dimerand trimerization (about equally). Both the magnitude of the nucleation rate and the qualitative description of the nucleation dynamics appear to be strongly affected. Inspecting the overall results, it seems quite possible that for certain saturation ratios multiple widelyspaced cluster sizes might make about the same contribution to the summation, in which case each would represent a rate-limiting bottleneck. Clearly this is related to magic number effects in cluster size distributions. This warrants further investigation, and may have interesting consequences for nucleation theory. ACKNOWLEDGMENTS This work was partially supported by the National Science Foundation (grant CTS-9520147) and by the Minnesota Supercomputer Institute. Helpful discussions with Dr. Paul Ziemann are gratefully acknowledged. REFERENCES Chelikowsky, J. R., Glassford, K. M. and Phillips, J. C. (1991) Phys. Rev. B 44, 1538. Gai, H., Thompson, D. L. and Raff, L. M. (1988) J. Chem. Phys. 88, 156. Girshick, S. L. and Chiu, C.-P. (1990) J. Chem. Phys. 93, 1273. Girshick, S. L., Rao, N. P. and Kelkar, M. (1996) J. Vac. Sci. Technol. A 14, in press. Hoare, M. R., Pal, P. and Wegener, P. P. (1980) J. Coll. Interface Sci. 75, 126. JANAF Thermochemical Tables, Suppl.1. (1985) J. Phys. Chem. Ref. Data 14. Martin, D. L., Raft, L. M. and Thompson, D. L. (1990) J. Chem. Phys. 92, 5311. McQuarrie, D. A. (1976) Statistical Mechanics. New York, Harper & Row. Raghavachari, K. and Rohlfing, C. M. (1988) J. Chem. Phys. 89, 2219.
BINARY CONTRIBUTION
TO NUCLEATION
RATES
C. F. C L E M E N T 1 and I. J. F O R D 2
1QuantiSci, Chiltern House, Henley-on-Thames, Oxon RG9 1AT, U.K. 2Department of Physics and Astronomy, University College, Gower St.,London WC1E 6BT, U.K. Abstract-We investigate the contribution of binary pairs to nucleation starting from rate equations for n-mer concentrations including growth from dimer collisions. Detailed balance for dimers in equilibrium determines decay rates in the same way as that for monomers. A steady-state current including dimer contributions is defined and we show how the technique of Becker and Doting (1935) can be extended to calculate the current. INTRODUCTION Rate equations form the most rigorous basis to nucleation theory (Clement 1992), but it has generally been assumed that only monomer addition to clusters contributes to a steady-state nucleation current. In this paper we show explicitly how to calculate a possible dimer contribution to the current by extending the method introduced by Becker and D0ring (1935) to determine the monomer current. GROWTH AND DECAY RATES Growth occurs from collisions between cluster i and cluster n leading to cluster i+n, where i= 1 and 2 for monomers and dimers, respectively. We denote the growth rates by CiCn~(i,n), where the ci are the concentrations of clusters of i monomers and/3(i,n) is the kinetic cross-section factor. Decay rates for clusters i+n to cluster i and cluster n are defined by ~,(n+i; i,n)Cn+ i. In equilibrium, the rates of all processes are equal to their inverse rates so that, (1)
7(n+i; i,n) CE,n+ i = 3(i,n) CEi CEn,
where the subscript E denotes equilibrium. The dimer decay rates can then be shown to be related to the monomer decay rates and the dimer, CE2, and monomer, c E l, concentrations in equilibrium by: -y(n+2; 2,n) = -y(n+2; 1,n+ 1) 7(n+ 1; 1,n) 3(2,n) CE2.
(2)
/3( 1,n+ 1)/3(1,n) c E 12 CURRENTS With actual values of Cn, the monomer and dimer currents are defined by:
=/3(1,n) c n c 1 - "),(n+1; 1,n) Cn+ 1,
(3)
J2(n) =/3(2,n) c n c 2 - "y(n+2; 1,n) Cn+2.
(4)
J 1(n)
It is then possible to write the set of rate equations for concentration changes as:
(5)
dcl/dt = - 2 J l ( | ) - ~i=2 Jl(i),
121
Clement and Ford
122 J2(3) I I
--
J 1
I
t J2(2) i
2
I
I
t
I
l
I
3
I
Jl(1)
Jl(2)
J2 (5)
)-
i J2(4) >
I 4 i
l,
i
)
I
"
t
)
J2(n- 1)1
J2(n+ 1)
I
->
17
av
1
I
5
6
I
Jl(3)Jl(4)
7
7
Jl(5)
8
Jl(6)
9
Jl(7)
n
n-1
7
i
J2(n+2)
J2(n)
J2 (7)
J2(6)
i
I
~)-
-}-
Jf(8)
9
I n+ 1 I
n+2
n+ 3
9
Jl(n-1) Jl(n) Jl(n+l) Jl(n+2)
Fig. 1. Monomer and dimer currents dc2/dt = J l ( 1 ) - J l ( 2 ) - 2J2(2) - ~i=3 J2(i),
(6)
dc3/dt = Jl(2) - J l ( 3 ) -
(7)
J2(3),
dcn/dt = J 1(n-l) - Jl(n) + J2(n-2) - J2(n), n > 3.
(8)
In steady-state nucleation, the populations, c 1 and c 2, are maintained at approximately constant values, as are concentrations c n up to the nucleation barrier. The steady-state nucleation rate, J, is given by the growth in the number of large clusters and, from the vanishing of sums of time derivatives given by these equations, or more simply from Fig. 1, can be written as: j-
Jl(2) + J2(2) = Jl(3) + J2(2) + J2(4) = J 1(4) + J2(3) + J2(4) = Jl(n) + J2(n-1) + J2(n).
(9)
These currents are the sums of those crossing the vertical dashed lines in Fig.1. We note that no asumption has been made that the monomer and dimer currents are separately constant as n increases. SOLUTION We now express J directly in terms of the concentrations, and solve the equations by a modified BeckerDOting procedure. J = b2c2-g3c3-G4c4 - B2c2+b3c3-g4c4-G5c5 = B3c3+b4c4-g5c5-G6c6 = Bn-I Cn-1 + bn Cn - gn+l Cn+l - Gn+2 Cn+2,
(10.1) (10.2) (10.3) (10.4)
where b n = 3(1,n) Cl + 3(2,n) c 2,
(11)
B n = 3(2,n) c 2,
(12)
Binary contribution to nucleation rates
123
gn = 7(3 1,2), n=3; = 7(n; 1,n-l) + 7(n; 2,n-2), n>3, G n = "t'(n; 2,n-2).
(13) (14)
The equations (10.1), (10.2), (10.3), (10.4) are multiplied by 1, A3, A4, An, respectively, and are added up to n=N above the barrier where we can neglect the decay terms containing CN+l and CN+ 2. The condition that the coefficients of c3...c N vanish in the sum is that the A i satisfy: B3 A4 + b3 A3 = g3,
(15.1)
B4 A5 + b4 A4- g4 A3 = G4,
(15.2)
Bn An+l + bn An- gn An-1- Gn An-2 = 0, n = 5,...N-1,
(~5.3)
bN AN- gN AN-1- GN AN-2 = 0.
(15.4)
Numerically, these equations can be solved efficiently by iteration and recurrence relations to specify J -
( 1 +
~i=3 to N Ai )-1 ( b2 + A3 B2 ) c2
(16)
CALCULATIONS To perform calculations it is necessary to specify monomer decay rates or equilibrium concentrations. Some preliminary calculations have been made using thermodynamic free-energy expressions: CEi = c 1 exp( - AFi/kT),
(17)
AF i = 0 ( i2/ 3-1 ) - (i- 1) In S,
(18)
where 0 = 7.84 and saturation S - 10. This produces a critical size of about 12. Full calculations allowing c 2 to change have not been performed and c 1 and c 2 had their relative equilibrium values of 1 and 0.1, respectively. An exact value of J was calculated from the full equations, but an approximate procedure, assuming monomer currents dominate, was used to determine the c i. The results are subject to confirmation, but their general nature is supported by analytic results obtained by approximating the equations: A. The ratio of the dimer current to the monomer current, Jl(n)/J2(n), is shown in Fig.2. For small n they are comparable, but the relative values at the barrier are close to the ratio of the dimer to monomer concentrations. B. The total current, J, is changed by only 1.3%, a reduction, from the monomer alone value. The reduction in J comes from a larger denominator in eq.(16), but it is somewhat surprising to find that allowing additional currents reduces the total current. There is apparently a bottleneck at the barrier which determines the total rate. More extensive calculations will be performed to justify an initial conclusion that, although dimer currents are large at low cluster size, their neglect in deriving the nucleation current is a good approximation because of their relative smallness at the barrier.
Clement and Ford
124
Ai multiplying factors
NOTATION G n decay coefficient (eq.(14))
B n growth coefficient (eq.(12))
gn decay coefficient (eq.(13))
b n growth coefficient (eq.(1 l))
J nucleation current
Cl monomer concentration
J 1(n) monomer current from n
c 2 dimer concentration
J2(n) dimer current from n
c n cluster concentration
S vapour saturation
tEn equilibrium concentration
T temperature
,y(n+2; 2,n) dimer decay
t time
factor from n+2
'~ff'n free energy for cluster n
13(1,n) monomer growth factor from n i3(2,n) dimer growth factor from n 3,(n+l; 1,n) monomer decay factor from n+ 1
0 factor in free energy REFERENCES Becker R. and Doting W. (1935) Am1. Physik 21 719. Clement, C. F. (1992) Nucleation and Atmospheric Aerosols. Proc. Thirteenth Int. Conf. Eds.N.Fukuta and P.E. Wagner, A. Deepak, Hampton, Virginia, pp.327-336.
c"
1.0--
1...
0
E
0.8
0c'0
E *-. 0.6 r-_
o
E
0.4-
.--
0.2
I
4
'
I
6
'
I
'
8
I
10
'
I
12
cluster size Fig. 2. Approximate ratio of dimer to monomer currents against cluster size in the nucleation model with critical size 12.
NUCLEATION IN ETHANOL VAPOR: THE EFFECT OF DIMER FORMATION J. HRUBY ~ and R. S T R E Y Max-Planck-|nstitut ftir Biophysikalische Chemie, Postfach 2841, 37018 G6ttingen, Germany ~Institute of Thermomechanics, Dolejgkova 5, CZ-18200 Prague, Czech Republic Abstract - We have performed extensive measurements of homogeneous nucleation rate J of ethanol vapor in argon as function of temperature T and activity a in the range of 2 x 105
where
~o- i, o ~ o~ ~ - ~ (~-~); c ~ - ~ [ ~ - ~); r
f
- 6,: + 5,:) ;
Substituting we
have
the
in
-
( 1 ) the v a l u e
+
~.7~
on
account
of
)
( 2
) -
( 5
)
relation
(6) where racy
coefficient and
contact
angle
depending
on
takes
following
the
the
third
degree
form:
r/R
of
accu-
Heterogeneous embryo radius on spherical substrate
"s(#,",..)-
Z3
+ 3o=. +~(-., .,~,,,,~-.,.
~~
147
+ ~.~j ""
~
~
/ "~
' ~P
1 - Wet Wall 2 - Dry Wall
Figure 7: Density profiles near the wall for an experiment involving 1-butanol and argon T e m p e r a t u r e effect: Figure 8 depicts the density profiles near the wet wall for l-propanol in helium at four different temperature ranges. For the lowest temperature range the density profile is nearly vertical and exhibits a minimum at the wall. At higher temperature ranges, the density profile decreases more strongly with reduced height resulting in a stable layering all the way to the wall. Thus, higher temperature ranges are favorable in avoiding this kind of convective flow. Density Profiles Near The Wall Temperature Study,~Wet Wall Case 1 -Propanol/Helium D/H = 7.5
~
4
i
'
~ - ff~ot ,, u.~ bar 2 - P l o t = 1.1 8 bar 3 - P l o t - 2 . 0 bar
Figure 9: Density profiles near the wall for varying total pressure for experiments involving I-propanol and helium. SUMMARY We have solved the two-dimensional mass and energy transport equations, modeling conditions at every point in the diffusion cloud chamber. Conditions at the wall were accounted for by a choice of appropriate boundary conditions. The results of these simulations are useful in understanding the operation and the limitations of a TDCC. Based upon these simulations, we conclude that using either a wet or dry wall does not affect the interior conditions of the TDCC. We find that the aspect ratio plays a crucial role, and for quantitative nucleation rate measurements it must be 7.5 or greater (provided the rate measurement is restricted to the center portion of the chamber) to ensure the validity of the assumption of onedimensional transport in the chamber. We also find that the possibility of convective instability is enhanced by low temperature and high pressure conditions that lead to density minima near the wall. ACKNOWLEDGMENTS
"o.~j" "~P'..~. "~" o . ~ - - ~ o9 ~
r
I 2 8 4
- TL=$02.BK, T U - 258.5K - T L - 3 1 5 . 6 K , TIJ-269.7K - T L - $ 2 4 . 6 K , TLI-276.gK - TL-$38.1 K, T U - 2 8 8 . S K
Figure 8: Density Profiles at the wall for varying temperatures for experiments involving l-propanol and helium Total pressure effect: Figure 9 shows the density profile near the dry wall for 1-propanol in helium at three different pressures. At pressures near ambient and in this low temperature range, the density gradient clearly ex-
The research described in this paper was supported in part by National Science Foundation grant No. CTS 8919847 and in part by the University of Rochester. ! Heist, Janjua, Ahmed: J.Phys.Chem. 1994, 98, 4443. Heist, Ahmed, Janjua: J.Phys.Chem. 1995, 99, 375. Bertelsmann, Stuczynski, Heist: J.Phys.Chem. (in press) 2 Bertelsmann, Heist: "Simulation of 2-Dimensional Transport and Study of Wall Effects in the Diffusion Cloud Chamber", submitted to J. Phys. Chem, 1996. 3 Katz, J.L., J.Chenz.Phys. 1970, 52, 4733
A C R I T I C A L L I N E L I M I T A T I O N OF E M B R Y O S LAPLASS'S PRESSURE M.P. A N I S I M O V a n d A.G. N A S I B U L I N
I n s t i t u t e of A t m o s p h e r i c Optics, Siberian B r a n c h of R u s s i a n A c a d e m y of Science, 1, A c a d e m i c h e s k i i Ave., 634055 T o m s k , Russia. E-mail:
[email protected] A b s t r a c t - F r o m a c o m m o n point of v i e w a v a p o r - gas s y s t e m is a solution w i t h partial solubility of components. W e are discussing t h e r a t e of n u c l e a t i o n of d i b u t y l p h t a l a t e (DBP) v a p o r in h e l i u m (He), c a r b o n dioxide (CO2) ( Anisimov, M.P. a n d V e r s h i n i n S.N., 1990 ), s u l f u r h e x a f l u o r i d e (SF6) a n d t h e n u c l e a t i o n of a g l y c e r i n - S F 6. at p r e s s u r e s 1.0-3.0 bar. Clear i n f l u e n c e of a b i n a r y s y s t e m critical line is s h o w n on a glycerin-SF6 s y s t e m n u c l e a t i o n rate. D e p e n d e n c i e s of a s a t u r a t i o n ratio l o g a r i t h m on n u c l e a t i o n t e m p e r a t u r e s at a fixed l o g a r i t h m of v a p o r n u c l e a t i o n rates u n d e r d i f f e r e n t total v a p o r - g a s p r e s s u r e s h a v e inflections at critical t e m p e r a t u r e of a t e s t e d system. T h e critical t e m p e r a t u r e of a p u r e s u l f u r h e x a f l u o r i d e is a varisimilar t e m p e r a t u r e limit of a s y s t e m critical t e m p e r a t u r e . It is possible t h e Laplass's p r e s s u r e of critical e m b r y o s is l i m i t e d b y critical p r e s s u r e s of a n u c l e a t e d system.
K e y w o r d s - V a p o r nucleation; B i n a r y system; Critical line
INTRODUCTION Efforts of m a n y scientists are d i r e c t e d to test of t h e nucleation theory. E x p e r i m e n t a l d e p e n d e n c i e s of v a p o r - gas n u c l e a t i o n r a t e are u s u a l l y c o m p a r e d w i t h a modification of a classic theory. Till this t i m e it is u n c l e a r t h e i n f l u e n c e of a c a r r i e r gas on a s u p e r s a t u r a t e d v a p o r nucleation. T h e effect of c a r r i e r - g a s on n u c l e a t i o n rates is e x a m i n e d in t h e o r e t i c a l a n d e x p e r i m e n t a l w o r k s (Anisimov, M.P. a n d V e r s h i n i n S.N. (1990), Katz, J.L., J.A. Fisk a n d V . C h a k a r o v (1992), W a g n e r , P.E., R.Strey, a n d Y.Viisanen (1992), Wilemski, G., B.E.Wyslouzil, M . G a u t h i e r , a n d M.B.Frish, (1992), Ford, I.J.(1992), Oxtoby,
203
204
Anisimov and Nasibulin D.W. a n d A . L a a k s o n e n (1995), Wilemski, G., B.E.Wyslouzil (1995), Wyslousil, B.E. et a1.(1994), Wyslousil, B.E., a n d G.Wilemski (1995), et al.). On e x a m p l e of b i n a r y solution an influence of a critical line on a v a p o r n u c l e a t i o n r a t e is discussed. This is close to a c a r t i e r - g a s p r e s s u r e problem. EXPERIMENTAL RESULTS AND DISCUSSION F r o m a c o m m o n point of view a vapor - gas s y s t e m is a solution w i t h partial solubility of components. As e x a m p l e of this s y s t e m h e r e w e are discussing t h e n u c l e a t i o n of d i b u t y l p h t a l a t e (DBP) v a p o r in h e l i u m (He), carbon dioxide (CO2) ( Anisimov, M.P. a n d V e r s h i n i n S.N., 1990;), s u l f u r h e x a f l u o r i d e (SF6) a n d t h e n u c l e a t i o n of a glycerin-SF6, at p r e s s u r e s 1.0-3.0 bar. It was u s e d for e x p e r i m e n t s a flow diffusion c h a m b e r like A n i s i m o v a n d C h e r e v k o (1985). H e l i u m has low critical p a r a m e t e r s . For CO 2 - D B P s y s t e m critical line is b e g u n in t h e CO 2 critical point w i t h a p r e s s u r e 73.87 b a r a n d a t e m p e r a t u r e 31.0 ~ In case of S F 6 - D B P s y s t e m critical line is b e g u n in point w i t h p r e s s u r e 37.59 b a r a n d t e m p e r a t u r e 45.5 ~ A n u c l e a t i o n r a t e g r a p h s of a DBP-CO2 t e m p e r a t u r e regions l o w e r a critical one classical t h e o r y at a critical t e m p e r a t u r e lines of n u c l e a t i o n rates vs a vapor classical. O n l y in region of t h e p r e s s u r e s small w a v e s of t h e nucleation rates.
s y s t e m h a v e a clear inflections in a n d are slightly p r o m i n e n t as in t h e above. In case of H e - D B P s y s t e m s u p e r s a t u r a t i o n are practically 1.5 b a r a n d h i g h e r w e can see t h e
Oscillatory r e l a x a t i o n of D B P vapors s p o n t a n e o u s nucleation at t e m p e r a t u r e n e a r 49 ~ w a s f o u n d (Fig.l). Nucleation r a t e oscillations are a l w a y s d e t e c t e d , b u t do not r e p e a t t h e m s e l v e s in d i f f e r e n t e x p e r i m e n t a l series. Oscillations a m p l i t u d e d e c r e a s e s r a p i d l y w h e n t h e t e m p e r a t u r e of n u c l e a t i o n is not in t h e vicinity of 49 ~ T h e d e p e n d e n c e of a v a p o r activity (a) on t h e n u c l e a t i o n t e m p e r a t u r e (T) at a fixed nucleation rate has a s h a r p c h a n g e at 49 ~ (Fig.2). This t e m p e r a t u r e is n e a r t h e critical t e m p e r a t u r e of S F 6 . It s e e m s t h a t a critical p r e s s u r e of t h e SF6 - D B P s y s t e m g r o w s w i t h i n c r e a s i n g of a n u c l e a t i o n temperature. W e h a v e got a t h e o r e t i c a l estimation of critical p r e s s u r e s of a H e - D B P , CO 2 - D B P a n d S F 6 - D B P b i n a r y s y s t e m s w h i c h are not h i g h e r 300 bar. Laplass's p r e s s u r e of critical e m b r y o s in a drop a p p r o x i m a t i o n is not l o w e r 600 b a r for t h e t e s t e d systems. It is a supercritical p r e s s u r e for t e s t e d
A critical line limitation of embryos Laplass's pressure systems. Wellknown a surface tension approaches to zero at a critical condition of a binary system (in our case) and it is impossible to get an embryo pressure higher than a critical condition one. Clear influence of a binary system critical line is shown on a glycerin-SF6 system (Fig.3). Dependencies of a saturation ratio logarithm (lgS) on nucleation t e m p e r a t u r e s (T) at a fixed logarithm of vapor nucleation rates (lgJ) u n d e r total vapor-gas pressures (P = 1; 2; 3 bar) have inflections at critical t e m p e r a t u r e of a tested system. The critical t e m p e r a t u r e of a pure sulfur hexafluoride is a varisimilar t e m p e r a t u r e limit of a system critical temperature.
We need to experimental the Laplass's pressure of a
SUMMARY AND CONCLUSIONS check traditional presumptions in the interpretation of results and revise of the theory postulates. It possible that pressure of the critical embryos is limited by a critical nucleated system.
A c k n o w l e d g m e n t s - Authors would like to acknowledge the Russian F u n d a m e n t a l Research F u n d for the grant 94-03-09947 and Russian Academy of Science for the State Scientific Scholarship for one of the Author. REFERENCES Anisimov, M.P. and A.G.Cherevko (1985). J.Aerosol Sci., V.16, P.97-107. Anisimov, M.P. and Vershinin S.N. (1990) J.Aerosol Science, V.21. Suppl. 1. P.11-18 Ford, I.J. (1992) in Nucleation and Atmospheric Aerosols, ed. by N.Fukuta and P.E.Wagner. Deepak, Hampton VA, P.39 Katz, J.L., J.A. Fisk and V.Chakarov (1992) in Nucleation and Atmospheric Aerosols, ed. by N.Fukuta and P.E.Wagner. Deepak, Hampton VA, P.11 Oxtoby, D.W. and A.Laaksonen (1995) J.Chem.Phys. (in press) Wagner, P.E., R.Strey, and Y.Viisanen (1992) in Nucleation and Atmospheric Aerosols, ed. by N.Fukuta and P.E.Wagner. Deepak, Hampton VA, P.27 Wilemski, G., B.E.Wyslouzil, M.Gauthier, and M.B.Frish (1992) in Nucleation and Atmospheric Aerosols, ed. by N.Fukuta and P.E.Wagner. Deepak, Hampton VA, P.23 Wilemski, G., B.E.Wyslouzil (1995) J.Chem.Phys. V.103. N.3. P.1127-1136 Wyslousil,B.E., G.Wilemski, M.G.Beals, and M.B.Frish (1994) Physics of Fluids. V.6. N.8.P.2845-2854 Wyslousil,B.E., and G.Wilemski (1995) J.Chem.Phys. V.103. N.3. P.1137-1151
205
Anisimov and Nasibulin
206 lgff" 6
_
~ o ~ 6"55
~0.
4 "
4
.0
,
,
.3 -3
,
2
"
"
, /
.
o
/
3s ~, o ""
O
9
o
q
l
"'l
~:-8
2' o ~
lg,a
_
_
Fig.1 An Experimental Nucleation Rate (J) on a DBP vap0r'-activity (a) of a D B P - S F 6 system at different nucleation t e m p e r a t u r e s (T).
lg a
2.0
1.8
e
45
50
Fig.2 DBP vapor activity on a nucleation t e m p e r a t u r e .
A critical line limitation of embryos Laplass's pressure
207
Ig3=3.2
1 . 9
igS
0
IgJ=l.5 1 . 7
r
1 . 6
P = i bar-
1 . 5
-.
i
'
1 . 4
35
_,. . . . . . . . . .
40
, ................
45
..c . . . . . . . . . . .
50
r--.~...~
"~..~
,
I ._ . . . . . . . . .
o --s...~.
!
t
~/'.~I~_ .A~L. . . . . . . .
55
L ..................
60
,c. . . . . . . . . . . . . . . . . . . . .
65
70 t.~
o[~
1.9 IgS
IgJ=3.2 1.7
I
1.6
1.5~ i. 4
gJ-
_ -4]--
P = 2 bar
I. . . . . . . . . . . . . . . . . . . . . :".;5
, .....................
40
--41-
! | _.,. ...................
45
~~
t
.. . ~ _ . ~ x i . ~
,. . . . .
50
55
I
. . . . . . .
, ........................
60
,_ ......................
65
,,
70 0
t: C 1.8
igJ=3.2
igJ:: i ~ : .
IgS 1.6
1.4
1.2
..................... L .................. _,_ .................. t ...~>
?';5
40
,4.5
,50
_
k
! ! !
! ! ~
.................... A ....................... l_ ...........................
,5,5
60
65
Fig.3 Dependencies of a glycerin vapor saturation ratio (S) on nucleation temperatures (T) (IgJ= 1.5; 3.2; P = i" 2; 3 bar)
/0
o 't, C
NUCLEATION
IN T H E P R E S E N C E
of A I R I O N S A N D A E R O S O L
PARTICLES
M. NOPPEL
Department of Environmental Physics, Tartu University, 18 Lqikooli Str., Tartu, EE2400, Estonia Abstract - It is shown that the replacement of difference equations of cluster size distribution, in the standard way in classical nucleation theory, by differential equation gives erroneous cluster size distribution in the case of cluster scavenging by an aerosol. The condition for the value of vapor pressure is found. The condition is that the effect of vapor distribution around an aerosol particle of given size on nucleation can be ignored. The cluster balance equations of ion-induced nucleation at low vapor pressure, in the case of ion recombination and in the presence of an aerosol are presented. Keywords - Nucleation; Aerosols; Ion-induced nucleation INTRODUCTION The effect of cluster scavenging by preexisting free molecule particles ((Shi et al., 1990), particles smaller in size than the free path of vapor molecules) on homogeneous nucleation has been studied by Friedlander (1983). McGraw et al. (1983) found a solution to the steady-state cluster balance equation of Friedlander (1983) in the form of a continued fraction, and evaluated the scavenging effect of aerosol for atmospheric conditions to be essential for compounds with saturation vapor pressure less than about 10 -4 Pa. Shi et al. (1990) transformed, as in classical nucleation theory, the difference equations of cluster balance into the differential equation and derived an anal3r expression for the rate of homogeneous nucleation in the presence of free molecule aerosol. Some discrepancies between the results of this work and the results of McGraw et al. (1983) are attributed to the presumable differences between solving the difference equations and differential equation. In the present report the influence of transformation of the difference equations into the differential equation on the resulting nucleation rate is studied. As the effect of larger particles on nucleation has drawn less attention, this effect is discussed in the case of compounds of low vapor pressure. Also the effect of air ions recombination and the scavenging effect of aerosol on ion-induced nucleation is considered. EFFECT OF TRANSFORMATION OF BALANCE EQUATIONS ON NUCLEATION RATE In most current theories of nucleation a supersaturated vapor is considered to be a mixture of monomeric molecules and molecular aggregates or clusters. Clusters are able to grow or evaporate by adding or losing monomers and clusters. Neglecting, as in classical theory, interaction between subcritical clusters, but including cluster scavenging by supercritical particles in free molecular regime, such a system is described by equations (Friedlander, 1983)
3 f g / O t = J g -Jg+, --Z~g, fli,gf, fg ,
2 10-6cm, both agree well. However, as R becomes smaller, they diverge significantly. At about the dimer position (dotted line), Tolman's value is a factor of about 10 larger than the present one. For long range interactions, f~ value increases due to interactions with molecules beyond the i
2
3
4
f~
0.500
0.866
1.225
2.000
2.449
0.204
0.3 54
0.500
0.816
1.000
0.058
O. 115
O. 179
0.433
1.000
= f~/f~
13
o~
....
o/ooo Table 1
The normalized surface molecular interaction factor, 1~., for the nearest neighbor interaction model.
Molecular theory of ultramicro clusters and nucleation. I
215
nearest neighbor, so that a factor must be multiplied to the o/ooo of the nearest neighbor interaction, f~,,n f=,Jr'
Y -
(~
and
= ~, lr
(s)
,,,'
where nn and lr stand for the nearest neighbor and long range interactions, respectively., Estimated yvalues are listed in Table 2. For example, for n = 6, (o2/o~o)t~ = 0.02, which is more realistic, and (o2/o~)u = (1/30)(oJo| n
4.5 .
Y = f~,,,/foo,l~_ Table 2
.
.
5.0 .
0.184
.
.
.
.
.
.
5.5 .
0.265
.
.
.
.
.
.
0.326
.
6.0
6.5
7.0
0.370
0.401
.
.
0.424 .
.
.
.
Ratio of f~ for the nearest neighbor interaction model to that for long range interaction model, as a function of n. For notations, see the text.
CONCLUSION Careful examination of thermodynamics revealed the basic interaction and misconception leading to o appearance on the surface of any curvature as well as the limitation of the so-called "thermodynamic method" developed only for gently curved surfaces. Two molecular methods were developed to estimate o, i.e., the Large Cluster Model (LCM) and the Small Cluster Model (SCM), applying a pair-wise attracting potential with hard sphere. The size dependence of o obtained by LCM agreed well with the Tolman equation in tendency. The SCM with the nearest neighbor interaction showed vanishing of o at i = I. An expression for o / o was obtained. The model gives o2/o~ = 0.06 which is 10% of the value of the Tolman equation. For the long range interaction with n = 6, o2/o= = 0.02. Based on these drastically small o values for ultramicro clusters, a new conceptual development and clarification of nucleation theory is anticipated. Acknowledgment. This study was supported by Division of Atmospheric Sciences, National Science Foundation, under Grant ATM-9112888. REFERENCES Abraham, F.F. (1974) Homogeneous Nucleation Theory, Academic Press, New York, 263 pp. Dufour, L. and Delay, R. (1963) Thermodynamics of Clouds, Academic Press, New York, 255 pp. Gibbs, J.W., (1961) Collected Works ofJ. W. Gibbs, Longmans Green and Company, New York, 434 pp. Ono, S. and Kondo, S. (1960) Molecular Theory of Surface Tension in Liquids, Springer-Verlag, Berlin, 280 pp. Pmppacher, M.R. and Klett, J.D. (1978) Microphysics of Clouds and Precipitation, D. Reidel, Dordrecht, Holland, 714 pp. Rigby, R., Smith, F.B., Wakeham, W.A. and Maitland, G.C. (1986) The Forces Between Molecules, Clarendon Press, Oxford, 232 pp.. Thomson, W. (1870) Proc. Roy. Soc. Edinburgh, 7, 63-68. Tolman, R.C. (1949) J. Chem. Phys. 17, 333-337. Wi~niewska, M. (1995) Surface Free Energy of Molecular Clusters, M.S. Thesis, Univ. of Utah, 66 pp.
NUCLEATION
FROM ATMOSPHERIC
FLUCTUATIONS
C. F. C L E M E N T 1 and R. G. HARRISON 2
1QuantiSci, Chiltern House, Henley-on-Thames, Oxon RG9 1AT, U.K. 2Department of Meteorology, University of Reading, Reading RG6 2AU, U.K. Abstract-The atmosphere is a very non-uniform medium with fluctuations appearing over a wide range of length and time scales in concentrations of condensible materials and temperatures. We describe their significance for nucleation, particularly the binary homogeneous nucleation of sulphuric acid. Small-scale fluctuations in wind, temperature and humidity were observed at Swiflerbank, North Holland, at a frequency of 18Hz. The physical origins of the observed correlations are deduced together with their implications for nucleation. INTRODUCTION Atmospheric temperatures, humidities and concentrations vary widely, and are subject to fluctuations arising from a variety of physical processes (Clement 1995). Mean values of these quantities will not specify the response of the atmosphere where the response is nonlinear, and this particularly applies to the process of nucleation. With fluctuations in saturations of order 10-2, Zapadinsky et al (1994) found only a small effect on the heterogenous nucleation of water vapour, but Easter and Peters (1994) showed that fluctuations could greatly affect sulphuric acid nucleation. Here, we examine the subject with reference to some observations of fluctuations made in the Netherlands.
RESPONSE TO FLUCTUATIONS For a vapour in the atmosphere, the main variables affecting its possible nucleation and condensation are the atmospheric pressure, p, the temperature T, and the vapour pressure, Pv, or density, Pv- They are each subject to fluctuations which may be correlated according to underlying physical processes, two of which we consider briefly here:Adiabatic changes. Rapid changes in pressure with no associated energy transfers induce correlated temperature changes: dp / p = 7 dT/( -y-1)T
(1)
where 7 is the specific heat ratio, 1.4 for air. From Bernouilli" s equation, pressure changes may be related to wind velocity, v, changes by:
dR = (1/2) p d(v2).
r
Heat and mass transfer. The main mechanisms are conduction and radiation for heat, and diffusion for vapour content. It is worth reiterating that convective motion at the constant pressure changes neither mass nor heat content in an air parcel, so that turbulence alone does not exchange heat and mass between different air parcels or between air and the earth's surface. There are a number of cases which can produce oppositely correlated fluctuations between heat and water vapour content, two of which are shown in Fig. 1.
216
Nucleation from atmosphericfluctuations cold
T < Ts
warm dry
unsaturated
T > Ts
air
air
Pv < PvE
heat
mass
p v < PvE
heat
IT
217
mass
+ T
Ts
warm water cold water Fig. 1. Production of air positively and negatively correlated between temperature and water vapour content. NUCLEATION Nucleation of water vapour in the atmosphere depends upon the saturation, S = pv/Pve(T), where Pve is the saturated vapour pressure, conveneiently specified by Bolton(1980) as Pve(T) = 6.112 exp[ 17.67 T(~
/ (T(~
(3)
+ 243.5 ) ] mb.
The sensitivity of the saturation to small changes in T and Pv, or equivalently Pv, is given by dS = dPv- 1 dPve(T ) d T S
Pv
Pve dT
4 o3
= dPvPv
(T(~
dT.
(4)
+ 243.5 )2
The binary homogeneous nucleation of sulphuric acid in the atmosphere which is very sensitive to, and increases with, increases in acid vapour concentrations and humidity, and decreases in temperature. The sensitivity of the nucleation rate to fluctuations of temperature and humidity in oceanic boundary has been investigated by Easter and Peters(1994), and evidence that fluctuations may be responsible for nucleation in large parts of the atmosphere has been deduced by Clement and Ford(1996). Easter and Peters(1994) found that the sensitivity of the nucleation rate to fluctuations depended strongly on the presence and nature of any correlations between temperature and humidity fluctuations. FLUCTUATION OBSERVATIONS Wind velocities, temperatures and absolute humidities (Pv) were observed at Swifierband, Holland, at 09.36h on 7 May, 1994. A total of 892 observations were made at a frequency of 18 Hz and, for temperature and humidity, the observation points were 20 cm apart. The mean values and standard deviations are shown in Table 1.The standard deviation in the square of the velocity gives a pressure change of 3.49 Pa from eq.(1) which, from eq.(2), corresponds to temperature fluctuations, AT / T = 9.84 10 -6 . This is two orders of magnitude less than observed so the origin of the temperature fluctuations lies in heat transfer processes. Correlations between the observables have been evaluated in the form; C (a,b)= Z i (a i - )(b i - )/[Z i (a i - )2Ei (b i - )2] 1/2
(5)
Clement and Harrison
218
Velocity square
Quantity q
u 2 + v2 + w 2
Temperature
Absolute humidity
T
pv
m 2 s-2
0C
g m -3
Mean < q > Standard deviation
17.436
15.46
14.086
Aq = 1/2
5.779
0.1962
0.2047
Aq / q
0.3314
6.798 10-4( T in K)
1.453 10 -2
Table 1. Quantities observed at Switterbrook.
We find C(T, Pv) -- 0.512, a significant positive correlation which can be seen in the fluctuations shown in Fig.2. This corresponds to heat and mass transfer processes, such as the first in Fig.l, where the differences between air and surface temperatures and water vapour contents are both the same sign. Somewhat surprisingly at first, we find significant negative correlations with velocity fluctuations, C(T,v 2)
-0.380 and C(pv,V2) = -0.471. These could, however, be qualitatively explained if we
postulate that higher velocities are leaving less time for conductive heat transfer and diffusional mass transfer from a warmer water surface. With the means and deviations of Table 1 and perfect correlation between Pv and T, the saturation deviation given by eq.(4) would be 0.0153 - 0.01191 - 0.00262. With the actual correlation of 0.512, this high cancellation does not take place and the standard deviation for dS/S is 0.0133. This is close to the figure of 0.01 used by Zapadinsky et al (1994) and, by inspection of Fig. 2, the fluctuation timescale is at least of the order of a nucleation ti,nescale of 0.1 s. Their conclusions for nucleation opposite to the "white noise" limit apply. To discover the consequences for sulphuric acid nucleation, we can refer to Fig. 6 of Easter and Peters (1994) with AT = 0.2 and their Aq = APv/p = 0.205/1.21=0.17 g kg -1. The increase in the binary homogeneous nucleation rate predicted for a correlation of 0.5 between the fluctuations lies between 1.1 and 1.5, a small increase compared to basic uncertainties in the rate. However, it is likely that measurements in other situations would reveal a larger increase due to fluctuations. To identify such situations, considerations of heat and mass transfer, like those shown in Fig. 1, are needed.
C correlation (eq.(5))
NOTATION T temperature u, v ,w velocity components
p pressure q observed quantity S saturation
v 2 square of total velocity Y ratio of specific heats p density
a, b correlated quantities (eq.(5))
Subscripts v refers to water vapour e refers to equilibrium
Nucleation from atmospheric fluctuations
219
REFERENCES Clement, C. F. (1995)J. Aerosol Sci. 26, $639-640. Clement. C. F. and Ford, I. J. ,(1996). The competition between aerosol growth and aerosol nucleation in the atmosphere, European Aerosol Conf. Easter, R. C. and Peters, L.K. (1994) Binary homogeneous nucleation: temperature and relative humidity fluctuations, nonlinearity, and aspects of new particle production in the atmosphere. J. Applied Meteorology 33 775-784. Zapadinsky, E. L., Sabelfield, K. K., Kulmala, M., Gorbunov, B. Z. and Rackimgulova, D. M. (1994) Monte Carlo simulations of heterogeneous nucleation on aerosol particles in the non-uniform media. J. Aerosol Sci. 25: S 101-S 102.
~176 E,
,
0.4
q,
,~T 1
'
,'
,I
"
i. ~
i
~
AT
o.o
,Tb_
.
-0.2
& e
lOs -O.8.
[
[
time
0.51
o.} 0.3
t~
. ~
'
0.2
APv g m -3
-0.6w
time
Fig. 2. Temperature (AT) and absolute humidity (Apv) fluctuations observed at Swifterbank during a 49.6 s time period.
A NEW SEMIPHENOMENOLOGICAL MODEL SIZE DEPENDENCE
FOR SURFACE
TENSION
P. S O P U C H + and A. SERVIDA 1=
f Institute of Thermomechanics, Academy of Sciences of the Czech Republic, Dolejskova 5, Prague :l:Dipartimento di Chimica e Chimica Industriale, Universitb. di Genova, Via Dodecaneso, 31 I- 16146 Genova, ITALY A b s t r a c t - In this communication, the bases of a new refined phenomenological model of cluster surface tension are presented. The model allows one to describe the dependence of the cluster surface tension with size. We also show that by augmenting the Lothe-Pound cluster free energy prescription with the new surface tension model provides an effective framework to represent in an unitary fashion the available experimental nucleation data for both the Becker-D6ring and Lothe-Pound fluids. Keywords - Surface tension, Cluster, Nucleation An essential part of the homogeneous nucleation models is the prescription of the cluster Gibbs free energy. Still nowadays, this represents a largely debated issue. There are two approaches. The first one the most well accepted relies on the capillarity approximation, while the second one makes use of the Lothe-Pound (LP) cluster model, which appears to the more consistent from the statistical point of view. In both cases, the surface contribution of the cluster free energy is evaluated making use of the macroscopic surface tension - assumption that may be unrealistic for clusters containing a few molecules. Within the framework of the Lothe-Pound theory, the Gibbs free energy of formation of a cluster consisting of i molecules is given by the expression: 2/3 AG i = Crsli - ikT In S + Ftr + Fro t - Fre p. (1) Here, a is the surface tension, S l is the equimolar surface of the monomer, T is the absolute temperature, S is the supersaturation, Ftr and Frot are the solid body translational and rotational Helmholtz free energies of the cluster, and Frep is the so called replacement free energy which is only -- -7kT. This theory gives good results for non-polar species using cy=croowhile it highly overestimates the nucleation rate for polar species. A possible explanation of the disagreement is that in reality for polar species c > (too. We have developed a simple model of the spherical cluster that really predicts a significant increase of the surface tension of small clusters of polar species (as opposite to the Tolman's correction). We use the classical approximation that a molecule moves in a potential field of its neighbours and the energy of the molecule is simply given by the sum of the internal energy of the molecule (we suppose that the one of the single molecule is not affected by its neighbours), of the kinetic energy like a solid body and of the potential energy in the field of near neighbours. The surface tension is an increase of the Helmholtz free energy as a consequence of the existence of the surface. Therefore, in our approximation cr is evaluated by the Helmholtz free energies of bulk and surface molecules. We use the approximation that the difference of potential energy of bonds strength highly dominates the difference of Helmholtz free energies of the molecule on the surface and in the bulk. Assuming that the bonds strength between the surface molecules and their neighbours are of the same magnitude like the ones between molecules in the bulk, noticing that the typical distance of neighbouring molecules in the cluster is -4~a (the second shortest distance in the structure fcc) where a-~/ml/Pliq, supposing that the typical area per one molecule on the surface is simply 2a 2 (the
220
A new semiphenomenological model for surface tension size dependence
221
square of the typical distance), and that the number of neighbours of the surface molecule divided by the number of neighbours in bulk, coefoo, equals approximately to 2/3 (Fig. 1.), we can write:
ooo =
u pot, bulk(coefoo_l) Upot,surf-Upot,bulk ('V~a) 2 = ("~a) 2 =-
Upot,bul k 3(.,~a)2 .
(2)
Assuming that the kinetic energy of one molecule in the liquid bulk, is approximately equal to that of single molecule in the gas phase, then Upot,bulk can be related to the heat of evaporation per one molecule. For liquids Ubulk----hliq, and hence: u pot,bulk = h liq - u kin,bulk - Uin = h g - h vap - u kin,bulk - Uin = (3) = Ukin,g + u i n
+kT-h
vap-
u kin,bulk - u in = kT -.- h v a p
Hence: hvap-kT ooo = 3(.V~a)2
(4)
9
The equation (4) satisfactorily relates fro,, to hvap for nonpolar species as opposite to the polar ones, for which eqn.(4) overestimates or,,,,. A possible explanation is that the surface dipoles have the tendency to orientate perpendicularly to the surface and therefore the free energy of the surface molecule is lowered. This may be described by the appearance of a peak part in the angle density distribution of bond potential energy w (Fig.2.). We assume that the spherical part of w. is equal to the corresponding value in the liquid bulk Wb= Upot,bulk/4rt.
21/2a
"~
~ ~ _~1~
C'
L~'
ufface
9
9
Figure 1.
Figure 2.
Figure 3.
Let us proceed to the case of a spherical cluster. The surface tension, defined for the equimolar surface is related to the number of the surface molecules by: se ~ = (frsurf-fbulk)n surf (5) Here, f denotes the Helmholtz free energy of a single molecule. However, a better estimate of nsurf could be obtained referring to the sphere connecting centers of surface molecules of radius re-a~2. Hence: cy
=
1_ a__~
frsurf - f b u l k
(6) 2r e (~/-2a)2 Approximating the Helmholtz free energy by the potential energy of bonds and using eqn. (2) we get: ~ _
( )2[
/
)1
1 a_.~ 1 1-~ ' urpot,surf - u ~ pot,surf 9 (7) 2re ( ' ~ a ) 2 ooo We assume that the peak part of w is not affected by curvature. From Figure 3. we see that the angle embracing the number of neighbours (proportional to the solid angle Ysol of the 'neighbours cone') divided by the number of the ones in the liquid bulk is:
Sopuch and Servida
222 ]too --71;+~
Ysol coef r = = 4n Hence: u pot,surf- u Hence:
1-cos~
pot,surf =
(8)
( c o e f r - c o e f ~ ) Upot,bulk.
(9)
2
(Y
1---~a 2r e
~c - o7,o = where: g
2
2
1+
hvap-kT
Pli___qq
ooo
m1 ]
cos
(10)
cos
2
'
_ _ Upot, bulk -
(11)
oooa 2
Determining the angle/3 from Fig. 3., we obtain: -4~ -a
2 ~c =
1+
1- a
cos
re-a//2 cos~'~176 2
2
(12)
Equation (12) shows that ~" parametrically depends only on density, surface tension of the plane surface, and on the evaporation heat. It is important to realize that the relationship (12) cannot be used for very small clusters because the peak part will be also affected by curvature. Using the Taylor expansion of eqn. (12) with respect to a 1/3 -1/3 a__ truncated to the second order and making use of the relation - (4n/3) i we obtain: re re -1/3 .-2/3 K i = 1 + otli + ot21 , (13) where ~
(7)~3
( 6 -1 )
2
--
(1
e
We have extended equation (13) with the third order term, obtaining: -1/3 -2/3 -1 •i = 1 + a l i + o~2i + o~3i
(15)
The coefficient a3 may be obtained from the condition that eqn. (1) gives the right value for r2 determined by comparison of the virial equation of state with the Mayer's conjecture. For AGi given by the Fisher's droplet model we have (Dillmann and Meier, 1991):
[(
kT Ps . . . . . . In ~c2 2~/3 ooos 1 q0 kT
21:-1
)3
q0
B ~-A
1 '
(16)
where ~: and qo are the constants of the Fisher's cluster model. Formally for the Lothe-Pound theory: x =-4
"
q0 =
2gmlkT h2
( /
24n m 1 Fre p ~ p-~m e x p \ - - ~ - - .
Combining equations (15) and (16) we obtain:
(17)
A new semiphenomenological model for surface tension size dependence
o~3 : 2
(
tx1 _o~_2 ~c2-1 21/~ 3 22/~
)fFt : 2
kT In - ( Psoo 2'2/3CooSl L \ q 0 k T
2-5
I3 q0NA
223
]t
-1
~ ct2 21/~3 2 ~ 3
. (18)
It can be shown that it is always a l > 0. For not very small clusters for which a2 can be neglected, eqn. (13) predicts ~"> 1. Figures 4 and 5 show the comparison between experimental data and the predictions of various nucleation models for water (expansion chamber) and chloroform (supersonic nozzle). DM stands for the Dillmann-Meier theory, B D for the Becker-D0ring theory, LP for the Lothe-Pound theory, and LPcor for the Lothe-Pound theory with our surface tension correction factor. We see, that LPcor reconciles the agreement between the Lothe-Pound theory and experimental results. LIST OF SYMBOLS (y (Yoo Sl T S AGi Ftr Frot Frep Uin ml Upot,surf Upot,bulk ukin,bulk Ukin,g hvap Ubulk hliq pIN a
se re nsurf ~(i) B
NA
qo i
Ps
Boltzmann's constant surface tension surface tension of a plane surface equimolar surface of the monomer absolute temperature supersaturation Gibbs free energy of formation of a cluster consisting of i molecules translational Helmholtz free energy of a cluster rotational Helmholtz free energy of a cluster replacement free energy internal energy of a molecule monomer mass potential energy of the single surface molecule potential energy of the single bulk molecule kinetic energy of the single molecule in the liquid bulk kinetic energy of the single molecule in the gas heat of evaporation per one molecule energy of the single bulk molecule enthalpy per one molecule in the liquid bulk liquid density the shortest distance in the fcc structure equimolar surface equimolar radius number of the surface molecules surface tension correction factor of a cluster (consisting of i molecules) second virial coefficient Avogadro number parameter in the expression for the surface tension correction factor parameter of the Fisher's cluster model parameter of the Fisher's cluster model number of molecules in a cluster equilibrium vapor pressure
J/K N/m N/m m2 K J J J J J kg J J J J J J J
kg/m 3 m
m2 m
mol/cm-3 mol-1
m-3 Pa
Sopuch and Servida
224
REFERENCES Dillmann, A. and Meier, G.E. (1991) J.Chem. Phys. 94, 3872. Miller, R.C. (1976) Ph.D. thesis, University of Missouri,, Rolla, MO. Anderson, R.J., Miller, R.C., Kassner, J.L.,Jr, and Hagen, D.E. (1980) J.Atmos.Sci. 37, 2509. Dawson, D.B., Wilson, E.J., Hill, P.G., and Russel, K.C., (1969) J. Chem. Phys. 51, 5389. 20
....
I ....
I ....
1 ....
J . l O 4cm.3s.1
I .... ..... -- --
15
~A
E .O (~ t_ 4_s m10
~N~N~a
9
0
....
220
I I
-----
LPcor Anderson
"
" ~
! ....
BD
LP
9
I ....
- DM
....
x\ " ~ ~
I ....
et
II
al!
I
Miller
~n..,,
I...,I,,..I 230
240
.... 250
!.,..I 260
.... 270
I,,,,I,,,,, 280
290
300
Temperature [K]
Figure 4. Comparison between various nucleation models and experimental measurements for H20.
j = l O~Scm -~s -1 0.01
QL. 0.001
190
200
210
220
230
240
Temperature [K]
Figure 5. Comparison between various nucleation models and experimental measurements for CHC13.
NUCLEATION
CONTROLLED
GROWTH
A . A . L U S H N I K O V +, M . K U L M A L A *
OF AEROSOL
PARTICLES
a n d H. A R S T I L A *
+Karpov Institute of Physical Chemistry 10, Vorontsovo pole, 103064 Moscow K-64, Russia, *Department of Physics, University of Helsinki, P.O. Box 9, FIN 00014 Helsinki, Finland A b s t r a c t - The condensational growth of aerosol particles initially formed by a nucleation process is considered within the theoretical approach proposed by Lushnikov ~: Kulmala, (1995) (referred hereafter to as LK) for treating source enhanced condensation processes. The molecules of condensing vapor are assumed to be able to form dimers, trimers etc. which evaporate very quickly if their masses are below certain value g = G, otherwise they are stable against evaporation and grow by joining the vapor molecules by one. The expression for the nucleation rate is derived. Special attention is given to the case G = 3. The probability for the dimer fragmentation as the function of temperature is found analytically. Keywords-
Nucleation, Condensation, Particle growth, Mass Spectra INTRODUCTION
Disperse particle formation and growth are of great interest in many branches of physics, physical chemistry and modern technology. Of special importance this problem is for the Physics of Aerosols (and Atmospheric Aerosols in particular). The aerosol formation processes are not yet completely understood despite numerous and of many years efforts of the scientists to attack this problem theoretically and experimentaly (see recent review by Laaksonen et al., (1995) and extensive citation therein). There are two directions of these attacks: 9 To find a proper parametrization of the particle formation rate, e.g. the spontaneous nucleation rate is attempted to be expressed in terms of the bulk surface tension, supersaturation ratio etc 4some additional parameters related to the embryo surface curvature and other size effects (see, e.g. Wilemski (1995) and references therein). This approach requires the Thermodynamics (implying a kind of equilibrium) to be exploited in deriving the final expression for the particle formation rate. The discrepancy by many decimal orders between such theories and experiments does not characterize this approach from the best side. Another, ab initio approach introduces into consideration the molecular interaction parameters (e.g. Ellerby et al (1991)). and tries to derive the thermodynamic properties of nucleating clusters from the "first principles". No unique description of the nucleation process is achieved, however. 9 To define the formation-growth kinetics in terms of the rates /3 and a of the elementary acts of a single molecule evaporation-condensation (see e.g. Nowakowski & Ruckenstein (1991) and references therein). In this case a formal kinetic approach is introduced allowing the formation process to be described in terms of particle size distribution evolving with time. The rates may be found from a theoretical consideration or defined experimentally. Of course, it is not easy and not yet fully done, but this problem is independent in a way from the solution of the kinetic equations describing the particle formation-growth process. Next, there is a hope that sufficiently developed spectra will reveal a kind of universality making thus the parametrization more easy and natural. Here we follow along the second route in studying the condensational formation of aerosol particles via nucleation stage. A spatially uniform system consisting of the carrier gas + condensible vapor + suspended disperse particles formed by the condensational process is considered. A source is assumed to provide the system with the vapor. The external conditions (temperature, pressure) support the nucleation of the vapor. The particles formed by the nucleation process grow then by condensation. The
225
226
L u s h n i k o v et al.
problem to solve is formulated as follows: to find the particle size distribution at a given moment of time, once initial conditions and the rates c~ and/3 are known. BASIC EQUATIONS Let us consider the condensing system consisting of the carrier gas and a source permanently producing molecules of a condensible substance (monomers) with the rate I(t) (the number of the monomers produced in a unit volume at a time). The mass of a condensing particle is measured in terms of the number g of the monomers comprising the growing particle. The key quantity to find is the mass spectrum %(t) the number of g-mers per unit volume at the time t. We assume: i. The molecules of the condensible substance are able to produce disperse particles along the scheme in which a monomer either attaches to the growing particle with the rate a9cl proportional to the m o n o m e r concentration cl(t) or escapes from its surface with the rate/39. ii. Evaporation is essential only for the particles whose mass is less than a given one denoted as G. This means that/3g = 0 at g _> G The LK consideration corresponds to G = 2 (dimers are stable). iii. The characteristic times of evaporation (/3~-1) are much shorter than that of the whole process ( o ( 1 / x / ~ I ) . This assumption allows us to consider the nucleation stage as a steady state process. Equations describing the particle formation-growth process look as follows: dc l oo G dt = I - Cl ~ % % + 2/32c2 + ~ /39c9 9=1
(1)
9--3
This equation describing time evolution of the monomer concentration contains now the contribution of evaporating g-mers. The second group of equations describes the nucleation stage which is assumed to be s t e a d y - s t a t e : 1 0 -- XO~lCl2 -- O~2CIC2 -- ~2C2 -}- ~3C 3 Z
(2) 0 - - O l g _ l C l C g _ 1 -- OlgClCg -- ]~9C9 ~- ~ g + 1 r 0 - - O Z G _ 2 C l C G _ 2 -- O ~ G _ I C l C G _ 1 -- ~ G _ l C G _ I
The nucleation stage finishes at g = G, and all subsequent equations describe the condensational growth of the particles formed by nucleation (g >__G): dcG
dt
= ZG(t) - a G c l c c
(3)
9
dcg d---t =
~
-
~162
where we introduced the nucleation rate (the rate of formation of G-mers)" ~G(t)
(4)
-- OIG-IC1CG-I
The strategy of solution of the whole problem includes the following three steps: i. to find the combination (~c-aClCC-1 entering the last equation of the set (2) as the function of cl. Since the time dependence of 2G enters only via cl(t) the former may be found on solving eq. (2)" ZG
1
~'---2- -- 1 + X 2 -}- Z2X 3 + . . . X 2 X 3 . . . X G _
1
(5)
Here we introduced: Z2(c1) = (alC~)/2 - - the rate of formation of dimers if the evaporation processes are forbidden and xa = f l g / ( % c i ) - c ; / c l with c~ = /~g/%. This expression appeared earlier in the
Nucleation controlled growth of aerosol particles
227
paper by Yang & Qiu (1986) ii.The "condensational part" of the program is treated in the same way as in LK. We introduce the new variable"
7 = jfOt cl(t')dt'
(6)
allowing one to cast eq. (3) into the form"
dcg = O~9-1Cg-1 -- 0~9C9 dr
(7)
The Green function gg(r) satisfying the following set of differential equations 1"
dga + saga = ~(r) dr
-
-
(s) d~ dr + aggg - a g - l g g - 1 formally solves eq. (7).
dr'
(9)
g9(7" -- TI)ZG(CI(TI)) CI(T--~
Cg(7") - -
o iii. Now it is possible to close eq. (1) for cl(t)"
de1
C1 ~
- - I -- G I " G ( C l )
-
)
r
R(r
-
Tt)~G(CI(Tt)) ~
0
where
oo
R(,)- ~ ~6~(~) 9-'G
dr'
c1(go
71go
7/2 99.0 %), and was the same ~ 40 used as in several other studies reported in this volume. ~= PRESSURE TRANSDUCER CALIBRATION The response of the pressure transducer (Kistler 9031, with charge amplifier 5011) is shown in Fig. 1. The calibration conditions were extended over a considerable Ap/p0 range, see insert. No systematic effect on temperature (in the range from 15 to 55 ~ nor total pressure
256
e,
20
,
l
2
Fig. 1
,
l
4
,
I
,
6
voltage change N
I
8
10
257
Effect of supersaturation, temperature and total pressure (in the range from 0.1 to 2 bar) was detected. As the insert shows, varying the conditions over these ranges results in only marginal deviation from the calibration factor previously determined in an extended calibration series. The magnitude of an expansion has no noticeble effect on the linearity of the pressure tranducer reading (see Fig. 1). Accordingly, the expansion pressure and also temperature are precisely determined. The latter is true, because the npentanol and argon form an ideal gas mixture to a high degree of accuracy, as calculations including higher order correction terms show. An increase of the nucleation rate with increasing magnitude of the expansion reported by Viisanen et al. (1993) is negligible for the constant small expansions with Ap/p0 0.3 considered here.
109
.....
n-pentanol .
.
.
.
.
108 $
107
10 n
2.70.
,"
IO s
5
10 15 supersaturation
Fig. 2
20
RESULTS NUCLEATION RATES VS. SUPERSA TURA TION AND TEMPERA TURE The nucleation rates vs. supersaturation for n-pentanol in argon are shown in Fig. 2. In order to perform the measurements we prepare a vapor carrier gas mixture with fixed ratio of vapor and carrier-gas in a separate receptacle. The variation of the supersaturation is achieved by varying the total pressure po n-pentanol before the expansion and along with it the vapor 30 pressure. The total pressures range from 0.5 to 0.7 ........ in0 bar. The pressure variation in this range is of subor2O dinate importance, as will be shown below. As can be seen, the variation of the nucleation rate over orders of magnitude is extremely steep. As noted preJ-,O'cmOs' viously by Hruby et al. (1996), this trend compares favorably with the self-consistent nucleation theory self-consistent theory (solid lines). The temperature dependencies can be ~8 o 7 compared in a so-called activity plot. In Fig. 3 we 9 this work 6 compile the present onset activities ao (- supersatu, or the average radius, < R >, of the clusters and its dependence on the initial supersaturation. Approximately, we may write the mass balance equation for the final state of independent growth in the form
(c-coo)-+
( c 0 - coo) = < n > N =
4~r < R >3 N, 3Ws
(4)
which is equivalent to < R >3=
(3~O,Coo \
co
47r ) ( N ) [ ( ~ - ~ - ~ ) - 1 ]
(5)
.
A substitution of the expression for N into eq.(5) yields, finally, 1/3
=C(T){
10/9
(c-~)-1}
[ln (c-~) ]
exp
I
B(T) I 31n
C(T) = 3a,
(
2(2p2)1/3
\ c~2 /
"
(6)
~
(7)
Number of clusters formed in nucleation-growth processes
An analysis of these equations shows that the number of clusters, N, increases with an increasing initial supersaturation while the average size, < R >, decreases. To this result a thermodynamic interpretation can be given as discussed earlier by Schmelzer and Ulbricht (1987) [10].
Acknowledgement The work was performed as a part of a project financed by the Bundesministerium f/ir Bildung, Wissenschaft, Forschung und Technologie (BMBF), Germany. One of the authors (V.V. Slezov) acknowledges also the financial support given by the SOROS foundation (Grant No. ISSEP-SPU042062).
References [1] Kaischew, R., Stranski, I. N., Z. Phys. Chem. A 170, 295 (1934) [2] Becket, R., Dhring, W., Ann. der Physik 24, 719 (1935) [3] Volmer, M.: Kinetik der Phasenbildung, Th. Steinkopf, Dresden, 1939 [4] Frenkel, Ya. I.: Kinetic Theory of Liquids, Oxford University Press, 1946 [5] Zeldovich, Ya. B., Soy. Phys. JETP 12, 325 (1942) [6] Zettlemoyer, A. C., (Ed.): Nucleation, Marcel Decker, New York, 1969; Nucleation Phenomena, Adv. Colloid Interface Science 7 (1977) [7] Gunton, J. D., San Miguel, M., Sahni, P. S.: The Dynamics of First-Order Phase Transitions In: Phase Transitions and Critical Phenomena, vol. 8, Eds. C. Domb, J. L. Lebowitz, Academic Press, London - New York, 1983 [8] Gutzow, I., Schmelzer, J.: The Vitreous State. Thermodynamics, Structure, Rheology, and Crystallization, Springer, 1995 [9] Binder, K., Stauffer, D., Adv. Phys. 25,343 (1976) [10] Schmelzer, J., Ulbricht, H., J. Colloid Interface Science 117, 325 (1987) [11] Slezov, V. V., Schmelzer, J., Tkatch, Ya. Y.: Number of Clusters Formed in Nucleation-Growth Processes: The Case of Kinetic Limited Growth, Phys. Rev. B, submitted for publication
267
THE ROLE
OF PARTICULATE
MATTER
(STRATOSPHERE
IN OZONE
PHOTOCHEMISTRY
AND TROPOSPHERE)
P.J. CRUTZEN Department of Atmospheric Chemistry, Max-Planck-Institute of Chemistry, P.O. Box 3060, D-55020 Mainz, Germany
Especially the discovery of the ozone more than ten years ago has clearly shown that heterogeneous reactions on aerosol can play a large role in ozone photochemistry. In the stratosphere low temperature reactions on nitric acid trihydrate (NAT), on water ice particles, or on particles consisting of ternary supercooled solutions of sulfuric acid, nitric acid and water, lead to rapid conversion of the "reservoir" species HC1 and C1ONO 2 into active chlorine (C1, C10, C1202) which act catalytically to rapidly destroy ozone. Because there is so much more particulate matter in the troposphere than in the stratosphere, it is most likely that they too play a significant role in tropospheric chemistry. Next follow some examples of processes that can be of major importance: * loss of boundary layer ozone in the Arctic during springtime: probably due to bromine activation reactions on snow and ice covered surface, or on ice particles. A high correlation of filterable bromine with low ozone had been discovered and more recent work has shown the existence of significant levels of BrO radicals. * the role of seasalt particles in the photochemistry of the marine boundary layer: It is quite possible that C1 and Br activation can occur by heterogeneous reactions on seasalt particles also in warmer environments, triggered a) in originally, continuentally polluted air masses via N 2 0 5 + X-
--,
XNO 2 + NO 3-
XNO 2 + hu
--,
X + NO 2
(X = C1, Br)
or
b) in NOx-pOor environments via HSO 5- + B r -
-,
SO4 = + H O B r
, followed in either case by the autocatalytic Br and BrO activating chain reactions HOBr + C1- + H +
--,
BrC1 + H20
268
The role of particulate matter in ozone photochemistry BrC1 + Br-
=
Br2C1-
=
Br 2 + hu
--,
2 Br
2(Br + 03
--,
BrO + 02)
2(BrO + HO 2
-,
HOBr + 02
net: 2 HO 2 + H + + 203 + Br- ---,
269
Br 2 + C1-
HOBr + 402 + H 2 0
The bromine activation (HOBr formation) next leads to catalytic 0 3 destruction via BrO + HO 2
--,
HOBr + 02
HOBr + hu
--,
OH + Br
OH+CO 2(+O2)
--,
CO 2 + H O 2
Br + 03
--,
BrO + 02
net: 03 + CO
--,
02 + CO 2
* Possible 03 (and NO) loss reactions in water droplets or on water coated ice particles
HO 2
--,
H + + 02-
02 - + O 3 + H +
---,
OH+202
02- + NO
--,
OONO-
--,
NO 3-
(?)
* Reactions of N205 on sulfate or other water containing aerosol:
N205 + H 2 0
---,
2HNO 3
, leading to NO x loss and diminishing 0 3 and OH concentration levels. * Interactions of DMS-derived sulfur compounds with seasalt in the marine boundary layer and reactions of SO2, H2SO4, NOx, N205 and HNO 3 on soil dust particles which remove these compounds from the gas phase. In the case of industrial SO2, the neglect of such heterogeneous reactions may well have led to overestimations of the climatic cooling effects of anthropogenic aerosols as many incorporation of sulfur in soil dust or seasalt will prevent the formation of new sunlight backscattefing sulfate particles.
270
Crutzen
REFERENCES Barrie, L.A. et al. (1988) Ozone destruction and photochemical reactions at polar sinrise in the lower Arctic atmosphere. Nature, 334, 138-141. Dentener, F. and Crutzen, P.J. (1993) Reaction of N20 5 on tropospheric aerosols: impact on the global distribution of NO x, 0 3 and OH. J. Geophys. Res., 98 (D4), 7149. Hausmann, M. and Platt, U. (1994). Spectroscopic measurements of bromine oxide and ozone in the high Arctic during polar sunrise experiment 1992. J Geophys. Res., 99, 25399-25413. Sander, R. and Crutzen, P.J. A model study indicating halogen activation in polluted air masses transported to the sea. J. Geophys. Res. (in press). Vogt, R., Crutzen, P.J. and Sander, R. A new mechanism for bromine and chlorine release from sea salt aerosol in the unpolluted marine boundary layer. Nature (submitted).
ICE NUCLEATION-
A REVIEW
GABOR VALI Department of Atmospheric Science, University of Wyoming Laramie, WY 82071, USA Abstract-Ice formation in cirrus has been successfully analysed by a number of authors in terms of homogeneous nucleation on dilute sulfuric acid aerosol: laboratory measurements were reconciled with nucleation rates derived from classical theory and field observations support the estimated temperature and humidity regimes at which ice formation is expected. Even so, there is need for further work on this topic. The origin of ice in lower tropospheric clouds is not resolved- it remains a question of great importance and in need of new efforts. In the realm of basic studies, the finding that monolayers and bacterial proteins can be very effective freezing nuclei opened new horizons and may also have implications for atmospheric processes. K e y w o r d s - Homogeneous freezing nucleation. Freezing nucleation by monolayers. Bacterial ice nuclei. Ice formation in clouds. INTRODUCTION The phase transitions of water to ice have crucial consequences in the atmosphere and in biological systems. Studies of the processes even with pure water face many obstacles, and the heterogeneous processes are yet more complex. Improvements in observational and computational methods are producing slow but significant progress. Three topics related to atmospheric ice nucleation will be covered in this review. Two of these - o n e concerning the lowest temperatures, the other those very close to 0~ - are areas of great scientific activity and significant advances. In the third, the traditional focus of atmospheric ice nucleation research, progress appears quite slow. UPPER TROPOSPHERIC AND POLAR STRATOSPHERIC CLOUDS Cirrus clouds have come into increasing scientific focus over the last decade or so, because of their role in the Earth's radiation budget and climate. Polar stratospheric clouds have been shown to be important participants in the complex chemistry leading to ozone depletion. These concerns brought about renewed interest in homogeneous freezing and in the supercooling of haze particles. New laboratory experiments and field observations were coupled with improvements in theoretical descriptions. There were no major conflicts in the data accumulated on homogeneous freezing over the fifty or so years since the first observations that the practical limit to supercooling of water in the laboratory is around -40~ The dependence of this limit on sample volume has been well demonstrated, and the rate of ice formation was successfully interpreted in terms of nucleation rates (Jsl). With rough values for many constants, especially for the interfacial energy between the ice embryo and water, the quasi-thermodynamic theory was reconciled in many slightly different forms with the empirical results. Reviews of work up to about 1990 are in G6tz et al. ( 1991 ), Pruppacher and Klett (1978) and numerous other texts. These laboratory and theoretical results were also in reasonable accord with observations of rapid glaciation in the tops of deep convective clouds and with the prevalence of ice in cirrus forming at temperatures near -40~ More recent work refined the situation in many respects. First, evidence was accumulated supporting the dominance of homogeneous freezing in the formation of cirrus at cold enough temperatures. Second,
271
272
Vali
further laboratory experiments extended somewhat the empirical data base. Third, the possibility was investigated that haze particles (concentrated solution droplets of subcritical size for growth at the prevailing humidities) may freeze directly. Fourth, revisions in the theory of homogeneous freezing yielded better agreement with observations. Fifth, model calculations of cirrus formation were greatly improved. Field evidence about ice formation in cirrus and in wave clouds at temperatures near the homogeneous freezing temperatures of water was obtained both by direct sampling from aircraft (Sassen and Dodd, 1988; Heymsfield and Miloshevich, 1993, 1995; Jensen et al., 1994, Str6m and Heitzenberg, 1994) and from lidar observations (Sassen and Dodd, 1988; Sassen, 1992). These papers are augmented by the growing body of publications reporting aircraft, lidar, radar and satellite observations of cirrus and other upper tropospheric clouds in general; these works continue to confirm that the great majority of these clouds is composed of ice crystals at concentrations of few tens to few hundreds per liter. While radiative properties of some cirrus indicate that higher numbers of small crystals may be present, this point is still not clarified. One of the direct observations of ice nucleation is the brief aircraft sounding reported by Sassen and Dodd (1988); the transition from droplets to ice crystals was observed within a height interval of 50 m in slowly ascending air at a temperature between-35.0 and-35.6~ These aircraft data were supported by a change in lidar depolarization ratio corresponding to the phase change. Heymsfield and Miloshevich (1993, 1995) report more detailed aircraft data that lead to very similar conclusions. Ice formation was observed to besparse at temperatures higher than -35~ and became very rapid at temperatures of-37~ and lower. Importantly, humidity and updraft measurements accompanied the particle observations. The, maximum relative humidities reached - limited by the uptake of vapor by growing droplets or crystals - were found to decrease with temperatures decreasing from near-34~ to-56~ Ice formation rates went from near-zero to very high values just past the points of maximum relative humidity, consistent with the onset of homogeneous nucleation as solution concentrations just decreased and droplet sizes increased sufficiently for the freezing rates to become significant. For the conditions of the observations, measurable or 'significant' rates were estimated as 0.01 cm -3 s-1. Calculated relative humidity peaks, based on observed droplet sizes at the onset of freezing, on laboratory values for Jsl and on assumed compositions of sulfuric acid for the nuclei of condensation, showed good agreement with observations. The representativeness of sampling from aircraft, the inaccuracies involved in obtaining truly Lagrangian observations and the fact that all the data cited come from the same season and location (October-December, Colorado) are some of the limitations that have to be borne in mind. Stimulating results from colder temperatures were reported by Knollenberg et al. (1993) and by Strtim and Heitzenberg (1994). Both sets of observations, in the tops of tropical cumulonimbus at temperatures near-80~ and in orographically induced cirrus at -55~ indicated the presence of small ice crystals (10..20 ~tm) in concentrations reaching 104 L -1. These concentrations are much higher than those usually observed for particles which can be definitely identified as crystals by direct capture or by imaging probes. According to Knollenberg et al., the most likely source of these crystals was the freezing of sulfuric acid droplets near the cloud-top temperatures, though the origin of crystals at lower levels in the clouds couldn't be ruled out. When combined with the results of Arnott et al. (1994) it is clear that the sizes of cirrus particles, at least at temperatures below -40~ tend to have a wide spread, tending to negative exponential distributions for sizes up to about 100 lam and more variable above that. Analyses of optical properties of cirrus have also pointed to the existence of large numbers of small crystals. It will be interesting to see to what extent the differences between these observations and those summarized in the preceding paragraph originate from different instrumental capabilities and from real variabilities in cloud properties. New laboratory data on homogeneous freezing was produced by DeMott and Rogers (1990). Using a large chamber subjected to slow expansion, they obtained measurements of Jsl in the temperature range -30...-38~ At temperatures above -34~ there was evidence for a significant contribution by heterogeneous nucleation. (The solute effect resulting from the CCN on which the cloud droplets form is negligible for droplet sizes of several p,m diameter in the chamber.) The rates deduced by DeMott and Rogers compare
Ice n u c l e a t i o n ~ A review
273
well with the rate estimated by Sassen and Dodd from field data: Jsl ~ 107 cm -3 s-1 at -36~ however the magnitudes of the experimental uncertainties are not well known and can be expected to be quite large owing to the indirect way that nucleation rate is deduced. In fact, it is clear that neither cloud chamber experiments, nor experiments with droplets suspended in other media (emulsions) can be freed from quite fundamental limitations in the determination of Jsl. The theory of homogeneous freezing has been reexamined in several of the papers already cited, and received a thorough revision by Pruppacher (1995). His formulation of the classical theory incorporates improved values for the thermodynamic constants, and incorporates singularities of these constants near -45~ (suggested by Angell (1982), and argued against by Bartell and Huang (1994)). However, the activation energy is estimated by Pruppacher from empirical values of Jsl, so that only consistency, not proof, has been achieved between theory and experiment. The classical nucleation theory of spherical embryos with bulk properties has been refined by the diffuse interface theory of Gr~in~isy (1993) and applied to the homogeneous freezing of water (Gr~imisy, 1995); he reports substantially improved agreement with observations. Observations of liquid droplets at temperatures lower than-40~ indicate some limit to generality of the role of homogeneous freezing in upper tropospheric clouds. Hallett and Lewis (1967) argued on the basis of optical phenomena, Sassen (1992) showed lidar depolarization data, and there are a number of reports of aircraft icing, all indicating that liquid droplets can exist in the atmosphere at temperatures below -40~ These observations can be interpreted as evidence for the existence of haze particles at those conditions, below the saturation needed for growth to droplets. The associated question is what may be the freezing temperatures of such haze droplets? That question is even more acute with respect to the freezing of sulfuric acid aerosol and of sulfuric acid-nitric acid-water mixtures in the stratosphere. Observations in polar stratospheric clouds (PSC's) showed clearly that transformations to solids take place at temperatures below 195 K, but the conditions and chemical composition of the clouds are still subjects of intense investigation (e.g. Tolbert, 1994). Earlier attempts to determine the nucleation temperatures of H2SO4/H20 solutions with relatively large sample sizes (Ohtake, 1993; Song, 1994) appear to have been influenced by heteronuclei, as evidenced by the small differences between the melting point curves (Gable et al., 1950) and the freezing points for acid concentrations to about 35% by weight. Beyer et al., (1994) working with smaller samples (5~tl) found approximately 22 ~ supercooling, still less than the accepted values for homogeneous nucleation. Recent data by Bertram et al. (1996), working with aerosol rather than bulk material, show much colder nucleation temperatures for this concentration range, falling along a smooth curve starting from the homogeneous nucleation temperature of about-35~ for pure water. This curve is very close to the theoretical estimates of Jensen et al. (1994) and of Larsen (1994) based on equations for homogeneous nucleation and on extrapolations of the relevant constants. The agreement between observations and theory (for the 0...35% concentration range) is fairly reassuring, though some reservations must be expressed. First, the formation of an ice embryo in a surrounding of both H2SO4 and H20 molecules in the liquid state should consider different kinetics than truly homogeneous nucleus formation. Second, solution effects for equilibrium conditions are used to estimate effects on homogeneous nucleation temperatures even though experiments such as Rasmussen and Mackenzie (1972) and Ganguly and Adiseshaiah (1992) showed that homogeneous nucleation temperatures are depressed by significantly larger amounts than the melting temperatures. The importance of this factor is emphasized by DeMott et al. (1994). Large uncertainties must therefore be acknowledged in the theory, even beyond the difficulty of assigning appropriate interfacial energy values. On the experimental side, the control of composition and of humidity, as well as the observation of the phase change are major difficulties. Further work on these topics is clearly warranted, including determinations of the chemical composition of the aerosol involved. While there is ample evidence for the prevalence of sulfuric acid aerosol in the upper troposphere, and the involvement of nitrates in the polar stratosphere, there are also disparate observations such as those of Hagen et al. (1995) and of Podzimek et al. (1995) of small fractions of soluble material in upper tropospheric aerosol in the size range 0.04...0.13 ~tm, and the existence of larger particles (to 5 ~tm diameter) in concentrations of the order 1
274
Vali
cm -3. It is worth noting that the formulation of detailed models of cirrus and PSC formation (Jensen et al., 1994; Larsen, 1994; Heymsfield and Miloshevich, 1995) and the parametric model of DeMott et al. (1994) offer the opportunity to evaluate theoretical and laboratory results in comparison with field data. For acid concentrations between 35% and 67% by weight, the solid phase is assumed to be sulfuric acid tetrahydrate (SAT). The large-volume experiments of Ohtake (1993) and of Song (1994) yielded solidification temperatures near 200 K, but with a large degree of scatter. Beyer et al. (1994) report similar temperatures but also show that the length of time to solidification is quite large, so that the freezing of small sulfuric acid aerosols of high concentrations is unlikely. Anthony et al. (1995) report on the basis of FTIR observations that submicron sulfate aerosol of 35...90% composition remained liquid for periods up to 3 hours when held at temperatures between 190 K and 230K. It thus appears that homogeneous nucleation in such sulfuric acid aerosol will only take place at temperatures < 190 K. For practical applications to PSC's, ternary systems involving water, sulfuric acid and nitric acid need to be considered but will not be discussed here as they are covered by the review paper of T. Peter in this volume. CLOUDS IN THE LOWER TROPOSPHERE The formation of ice in lower tropospheric clouds, between the temperatures of the melting point and of homogeneous nucleation, remains a question of great importance for understanding the global climate and water balance yet progress is quite slow in this area. The fundamental problem is that the origins of ice particles are understood only partially and in very rough terms. Predictive capabilities are limited by the complexities of tropospheric aerosol, by inadequate theoretical formulations and by the lack of proven instrumentation for ice nucleus measurements. Interpretations of field observations are hindered by the direct connections between ice formation and cloud dynamics and by the existence of secondary ice generation processes. Recognition of these limitations resulted from the period around the 1970's which saw intense laboratory experimentation as well as numerous field studies utilizing aircraft equipped with ice particle probes. The major findings of that period, and progress since then, led to the recognition that ice particle concentrations in clouds are at times much higher than measured ice nucleus concentrations, that some of that discrepancy is due to secondary processes such as the Hallett-Mossop process, and that ice nucleation can proceed via different pathways and from a variety of natural nucleating substances. The current situation can be best characterized as one focussed on searching for solutions to overcome or bypass one or other of problems which stand in the way of making the findings more precise, more specific to given cloud systems and instances. Inherent to this search is the openness to new ideas. Two concepts have been put forward that may lead to new findings. Braham (1986) and Beard (1992) call attention to the fact that in warm-based convective clouds the freezing of rain or drizzle drops grown by coalescence seems to be a key element of ice initiation and development. Field observations of Rangno and Hobbs (1991) in maritime clouds appear to support the role of drop freezing in what they term the first stage of ice enhancement. In contrast, for continental clouds Rangno and Hobbs (1994) show evidence for possible linkages between cloud depth and the width of the droplet spectrum at cloud top and ice initiation. Common to all of these scenarios is some connection between events traditionally considered part of the warm-rain process and ice development. Possible mechanisms that lead to preferential freezing of the larger cloud droplets or rain drops are condensation on large condensation-freezing nuclei or the capture of large freezing nuclei by the drops. These processes have not yet been properly observed. Also, the relative sparsity in precipitation samples of freezing nuclei active at temperatures above-8~ would have to be reconciled with the idea of drop freezing in clouds by demonstrating that the activity is lost with time. The other suggestion that is motivating further investigations is that the residues of evaporating cloud droplets have enhanced ice nucleating abilities. Beard (1992) explores this idea and also links it with the possibility that such residues might be the contact-freezing nuclei discussed in the preceding paragraph. Laboratory evidence for this process is claimed by Rosinski and Morgan (1991).
Ice nucleation ~ A review
275
The measurement of ice nuclei with high spatial and temporal resolution as well as with indication of the mode of action remains an elusive goal. Some progress may emerge from the continuous-flow diffusion chamber design of Rogers (1994). In laboratory studies of silver iodide and related aerosols, DeMott (1995) showed the that the relative contributions of different nucleation modes depend on the aerosol type, temperature and, in some cases, on the specific history of when the aerosol is introduced into the cloud. These data opened the opportunity for more detailed comparisons with the results of cloud seeding (Meyer et al., 1995) than has been possible earlier. In general, modeling of the initiation of ice is done in ever greater detail, though the input assumptions remain roughly defined (e.g. Meyer et al., 1992). On the other hand, Baker (1991) performed calculations to show that a process often considered in ice nucleation - activation by high supersaturations with respect to w a t e r - is unlikely to be significant in clouds. ORGANIC ICE NUCLEI While monitoring of atmospheric ice nuclei, and observations of their activity in creating ice particles within clouds, are the principal concerns for establishing the origins of ice in clouds, the questions what substances and what mechanisms make efficient heteronuclei are clearly complementary to the primary foci. Important new results in these areas are summarized in the following paragraphs. Epitaxial fit between ice and the nucleating substrate is a well known factor favoring nucleation. That an alcohol monolayer could provide such a substrate and nucleate ice at temperatures near- 1~ was a startling discovery (Gavish et al., 1990). In fact many details of the monolayer structure, of the monolayer-ice interface and of the size of the ice embryo has now been elucidated by Popovitz-Biro et al. (1994) and Majewski et al. (1994). The initial work reported in 1990 demonstrated that 2D crystals of aliphatic alcohol monolayers whose lattice dimensions are close to that of ice would be efficient ice nuclei. The alcohol series CnH2n+IOH for n= 16...31, in the form of monolayers spread over water drops, produced a double series of nucleation temperatures for as n increased. For each series, nucleation temperatures rise toward higher values from near -14~ the freezing temperature found for the smallest n-values. For even values of n the highest nucleation temperatures reached -8~ for n>22, and for odd values of n the nucleation temperatures gradually rose to near-1 ~ for n=31. In comparison, carboxylic acids of similar chain lengths produced ice nucleation at temperatures around-16~ The lattice match, as determined by grazing-incidence X-ray diffraction (GID), was better for the alcohols than for the acids confirming the role of epitaxy. In the more recent reports, measurements with D20 covered with the aliphatic alcohol monolayers confirmed the findings with H20, showing the same series of nucleation temperatures but shifted toward higher temperatures. Monolayers of mixtures of two different alcohols, which produce poorly ordered surfaces, led to lower nucleation temperatures than either of the components separately; this confirms the role of structural fit to ice. This contrasts with the situation when up to 50% of a fluorocarbon alcohol is mixed with the n=20 hydrocarbon alcohol: no change in nucleation temperatures was found indicating that these two alcohols form separate crystalline domains. The influence of the chain length is interpreted as governing the lateral coherence length, and the tilt angle of the chain with respect to the b-axis, in all leading to better lattice match with increasing chain length. Perhaps the most difficult point to clarify is the reason for the difference between the even and odd members of the series, for example why the two longest members of the series, n=30 and n=31 yield nucleation temperatures differing by 7~ There is no clear evidence for differences in packing arrangements, so the hypothesis is that the orientation of the (CH2OH) head groups, and possibly rearrangement of this group, account for the differences in nucleating ability. This notion is also supported by experiments with alcohols in which ester or amide groups have been inserted mid--chain; the dependence of nucleation temperatures on overall chain length and parity was overridden by a dependence of the parity of the hydrocarbon fragment connecting the OH head group to the introduced functional group.
276
Vali
Examinations, using GID, of the monolayers just prior to and immediately following ice nucleation, as well as GID patterns from the ice itself allowed Majewski et al. (1994) to set an upper threshold for the critical size of the ice embryos at temperatures very near 0~ A coherence length of 25 ,~ was found for ice formed under the alcohol monolayer (which in itself has much larger coherence length) so that the critical size of nuclei is taken to be about 20 ]k corresponding to about 50 water molecules. [This is a surprisingly small critical size, comparable to that estimated to be necessary for homogeneous nucleation near-40~ In addition to the critical-size estimates, the GID observations showed that the monolayer undergoes a gradual change as the temperature is lowered from 6~ to 1~ but retains its structure through nucleation, growth and subsequent melting of the ice underneath it. The nucleated ice crystals were shown to have their c-axes perpendicular to the interface. Since binding of an ice layer to the monolayer would impose proton ordering, the question arises, and remains unanswered at this point, how far this proton ordering might extend into the ice. At least a partial, independent confirmation of the nucleating abilities of the aliphatic alcohol monolayers is given by Davey et al. (1994). Working with the relatively inactive n=30 member of the series, they found that nucleation is almost immediate with that monolayer at-8~ and that at higher temperatures, to -4~ nucleation will take place at with gradually increasing delay times (to near 12 h at -4~ Unfortunately, the nucleation temperatures observed in their apparatus with the monolayer differed only by few degrees from those obtained in the control experiments, so that the presence of other sources of nucleation can't be ruled out. The difference noted many years ago between threshold temperatures for freezing nucleation of chiral or racemic varieties of amino acids was confirmed anew by Gavish et al. (1992). These materials have very poor lattice fit to ice and do not differ in this respect from one another. Thus the observations require other interpretation. By noting that higher nucleation temperatures are found for those varieties that have polar axes, independently whether that occurs for the chiral-resolved or racemic variety, the authors propose that electric fields within cracks of the crystals raise the nucleation temperatures. Visual observations confirmed that cracks are preferred locations for deposition nucleation at-15~ Macroscopic electric fields, and electric charges on particles have been suggested before as factors in ice nucleation, but solid evidence has been difficult to produce; whether the suggestion for this microscopic electric-field effect will be more amenable to confirmation will remain to be seen. Bacterial ice nuclei were discovered in the 1970's. Their activity as freezing nuclei at temperatures as high as-2~ raised a great deal of interest in identifying the active agents. Practical applications were identified for bacterial nucleants as initiators of freezing, as well as for the elimination of these bacteria and consequent delaying or elimination of freezing in natural systems. Fungal species of similar activity were also found by the late 1980's. Most of the material relating these biological ice nucleants is now assembled in a single volume (Lee, Warren and Gusta, 1995) so readers are referred to that for details. What is missing from the compendium, for lack of any systematic attack of the issue, is work on the possible atmospheric role of these nucleants. For evolutionary reasons still not known, bacteria of a handful of species (out of the millions in existence) developed the ability to synthesize ice nucleating proteins in their outer membrane. In contact with supercooled water they initiate freezing at temperatures close to 0~ The fraction of individuals within a population of bacteria (either natural or cultured) which nucleate ice at a given temperature is quite variable and often quite small. Also, the nucleating material is a small fraction of the bacterial membrane. These facts made identification and isolation of the active agent quite difficult. It was accomplished by identifying the DNA segment responsible for expression of nucleating ability, confirming this by splicing this gene into otherwise non-active species and then deducing the protein structure that the gene elicits. Presence of the protein in the outer membrane was confirmed and visualized by antibody attachment; the protein forms clusters of a large range of sizes in abundances related to the activity of the sample. The protein molecules
Ice nucleation ~ A review
277
are 150...200 kDa size and are characterized by repetitive segments of nucleotides. Quantitative comparisons between protein amount per cell, numbers of clusters and nucleating ability led to the realization that activity is a function not only of the amount of protein produced but also of factors governing its aggregation. What these factors are is not yet established, nor is the structure of the nucleating site. Gamma ray deactivation analysis led to an estimation of the sizes of the nucleating units: at-13~ a single protein molecule may be sufficient while a t - 3 ~ a cluster of = 104 molecules is indicated. Perhaps the broadest impact that studies of organic ice nuclei have is that they produced new, quantitative methods for the characterization of the nucleation process and of the substrate surface. At the same time, ice nucleation came to the attention of scientists in diverse fields of research, with many of these scientists using ice nucleation as an analytical tool for the elucidation of genetic, plant pathological and many other problems. Implications for atmospheric ice nucleation are speculative - there is little known about the organic components of atmospheric aerosol in general, and it seems certain that research to improve that situation will be quite demanding considering the tiny amounts of organic material that may be involved in producing ice nucleation. Assessments of particles of biological origin in air and in precipitation (e.g. Matthias-Maser and Jaenicke, 1992; Casareto et al., 1996) indicate that the search should be continued. REFERENCES Angell, C. A., 1982: Supercooled water, in "Water- a comprehensive treatise, Vol. 7", Franks, E (Ed.), Plenum Press, pp. 1-81. Anthony, S. E., R. T. Tisdale, R. S. Disselkamp, M. A. Tolbert and J. C. Wilson, 1995: FI'IR studies of low temperature sulfuric acid aerosols. Geophys. Res. Lett., 22, 1105-1108. Arnott, W. P., Y. Y. Dong, J. Hallett and M. R. Poellot, 1994: Role of small ice crystals in radiative properties of cirrus- a case study, FIRE II, November 22, 1991. J. Geophys. Res. (Atmosph.), 99, 1371-1381. Baker, B. A., 1991: On the nucleation of ice in highly supersaturated regions of clouds. J. Atmos. Sci., 48, 1904-1907. Bartell, L. S. and J. Huang, 1994: Supercooling of water below the anomalous range near 226 K. J. Phys. Chem., 98, 7455-7457. Beard, K. V., 1992: Ice initiation in warm-base convective clouds: an assessment of microphysical mechanisms. Atmos. Res., 28, 125-152. Bertram, A. K., D. D. Patterson and J. J. Sloan, 1996: Mechanisms and temperatures for the freezing of sulfuric acid aerosols measured by FTIR extinction spectroscopy. J. Phys. Chem., 100, 2376-2383. Beyer, K. D., S. W. Seago, H. Y. Chang and M. J. Molina, 1994: Composition and freezing of aqueous H2SO4/HNO3 solutions under polar stratospheric conditions. Geophys. Res. Lett., 21, 871-874. Braham Jr., R. R., 1986: Coalescence-freezing precipitation mechanism. 10th Conf. on Planned and Inadvertent Weather Modification, Amer. Meteor. Soc. 142-145. Casareto, B. E., Y. Suzuki, K. Okada and M. Morita, 1996: Biological micro-particles in rain water. Geophys. Res. Lett., 23, 173-176. Davey, R. J., S. J. Maginn, R. B. Steventon, J. M. Ellery, A. V. Murrell, J. Booth, A. D. Godwin and J. E. Rout, 1994: Nucleation of crystals under Langmuir monolayers - kinetic and morphological data for the nucleation of ice. Langmuir, 10, 1673-1675. Demott, P. J., 1995: Quantitative descriptions of ice formation mechanisms of silver iodide-type aerosols. Atmosph. Res., 38, 63-99. DeMott, P. J. and D. C. Rogers, 1990: Freezing nucleation rates of dilute solution droplets measured between -30 ~ and -40~ in laboratory simulations of natural clouds. J. Atmos. Sci., 47, 1056-1064.
278
Vali
Demott, P. J., M. P. Meyers and W. R. Cotton, 1994: Parameterization and impact of ice initiation processes relevant to numerical model simulations of cirrus clouds. J. Atmos. Sci., 51, 77-90. Gable, C. M., H. F. Betz and S. H. Maron, 1950: Phase equilibria of the system sulfur trioxide-water. J. Amer. Chem. Soc., 72, 1445-1448. Ganguly, S. and K. S. Adiseshaiah, 1992: Ice nucleation in emulsified aqueous salt solutions- a differential scanning calorimetry study. Colloids and Surfaces, 66, 105-111. Gavish, M., R. Popovitzbiro, M. Lahav and L. Leiserowitz, 1990: Ice nucleation by alcohols arranged in monolayers at the surface of water drops. Science, 250, 973-975. Gavish, M., J. L. Wang, M. Eisenstein, M. Lahav and L. Leiserowitz, 1992: The role of crystal polarity in alpha-amino acid crystals for induced nucleation of ice. Science, 256, 815-818. Gtitz, G., M6sz~iros, E., and Vali, G., 1991" Atmospheric Particles andNuclei, Akad6miai Kiad6, Budapest. Gr~imisy, L., 1993: Diffuse interface theory of nucleation. J. Non-Crystalline Solids, 162, 301-303. Gr~.nS.sy, L., 1995: Diffuse interface analysis of ice nucleation in undercooled water. J. Phys. Chem., 99, 14182-14187. Hagen, D. E., J. Podzimek and M. B. Trueblood, 1995: Upper-tropospheric aerosol sampled during project FIRE IFO II. J. Atmos. Sci., 52, 4196--4209. Hallett, J. and R. E. J. Lewis, 1967: Mother of pearl clouds. Weather, 22, 56--65. Heymsfield, A. J. and L. M. Miloshevich, 1993: Homogeneous ice nucleation and supercooled liquid water in orographic wave clouds. J. Atmos. Sci., 50, 2335-2353. Heymsfield, A. J. and L. M. Miloshevich, 1995: Relative humidity and temperature influences on cirrus formation and evolution: Observations from wave clouds and FIRE II. J. Atmos. Sci., 52, 4302-4326. Jensen, E. J., O. B. Toon, D. L. Westphal, S. Kinne and A. J. Heysmfield, 1994: Microphysical modeling of cirrus .1. Comparison with 1986 FIRE IFO measurements. J. Geophys. Res. (Atmosph.), 99, 10421-10442. Knollenberg, R. G., K. Kelly and J. C. Wilson, 1993: Measurements of high number densities of ice crystals in the tops of tropical cumulonimbus. J. Geophys. Res., 98, 8639-8664. Larsen, N., 1994: The impact of freezing of sulfate aerosols on the formation of polar stratospheric clouds. Geophys. Res. Lett., 21,425-428. Lee, R. E. Jr., Warren, G. J., and Gusta, L. V., 1995: Biological Ice Nucleation and Its Applications, APS Press, The American Phytopathological Society, St. Paul, Minnesota. Majewski, J., R. Popovitzbiro, K. Kjaer, J. Alsnielsen, M. Lahav and L. Leiserowitz, 1994: Toward a determination of the critical size of ice nuclei - a demonstration by grazing incidence X-ray diffraction of epitaxial growth of ice under the C31H63OH alcohol monolayer. J. Phys. Chem., 98, 4087-4093. Matthias-Maser, S. and R. Jaenicke, 1992: Identification and size distribution of biological aerosol particles with radius > 0.2 um., in "Nucleation and Atmospheric Aerosols", Edited by N. Fukuta and P.E. Wagner, A. Deepak Publ., pp.413-416. Meyers, M. P., P. J. Demott and W. R. Cotton, 1992: New primary ice-nucleation parameterizations in an explicit cloud model. J. Appl. Meteor., 31,708-721. Meyers, M. P., P. J. Demott and W. R. Cotton, 1995: A comparison of seeded and nonseeded orographic cloud simulations with an explicit cloud model. J. Appl. Meteor, 34, 834-846. Ohtake, T., 1993: Freezing points of H2SO4 aqueous solutions and formation of stratospheric ice clouds. Tellus B - Chem. Phys. Meteor, 45, 138-144.
Ice nucleation B A review
279
Podzimek, J., D. E. Hagen and E. Robb, 1995: Large aerosol particles in cirrus type clouds.Atmos. Res., 38, 263-282. Popovitzbiro, R., J. L. Wang, J. Majewski, E. Shavit, L. Leiserowitz and M. Lahav, 1994: Induced freezing of supercooled water into ice by self-assembled crystalline monolayers of amphiphilic alcohols at the air-water interface. J. Amer. Chem. Soc., 116, 1179-1191. Pruppacher, H. R., 1995: A new look at homogeneous ice nucleation in supercooled water drops. J. Atmos. Sci., 52, 1924-1933. Pruppacher, H. R. and Klett, J. D., 1978: Microphysics of clouds and precipitation, D. Reidel Publ. Co., Dordrecht, Holland. Rangno, A. L. and R V. Hobbs, 1991: Ice particle concentrations and precipitation development in small polar maritime cumuliform clouds. Quart. J. Roy. Meteor Soc., 117, 207-241. Rangno, A. L. and E V. Hobbs, 1994: Ice particle concentrations and precipitation development in small continental cumuliform clouds. Quart. J. Roy. Meteor Soc., 120, 573-601. Rasmussen, D. H. and A.P. MacKenzie, 1972: Effect of solute ice-solution interfacial free energy; calculation from measured homogeneous nucleation temperatures, in "Water structure at the water-polymer interface", Edited by H. H. G. Jelinek, 126-145. Rogers, D. C., 1994: Detecting ice nuclei with a continuous-flow diffusion chamber-some exploratory tests of instrument response. J. Atmos. Ocean. Tech., 11, 1042-1047. Rosinski, J. and G. Morgan, 1991: Cloud condensation nuclei as source of ice-forming nuclei in clouds. J. Aerosol Sci., 22, 123-133. Sassen, K., 1992: Evidence for liquid-phase cirrus cloud formation from volcanic aerosols -climatic implications. Science, 257, 516-519. Sassen, K. and G. C. Dodd, 1988: Homogeneous nucleation rate for highly supercooled cirrus cloud droplets. J. Atmos. Sci., 45, 1357-1369. Song, N. H., 1994: Freezing temperatures of H2SO4/HNO3/HO2 mixtures: Implications for polar stratospheric clouds. Geophys. Res. Lett., 21, 2709-2712. Str6m, J. and J. Heintzenberg, 1994: Water vapor, condensed water, and crystal concentration in orographically influenced cirrus clouds. J. Atmos. Sci., 51, 2368-2383. Tolbert, M. A., 1994: Sulfate aerosols and polar stratospheric cloud formation. Science, 264, 527-528.
FORMATION MECHANISMS OF POLAR STRATOSPHERIC CLOUDS THOMAS PETER Max-Planck-Institut ffir Chemie, Mainz, Germany A b s t r a c t - The physical chemistry of polar stratospheric clouds is far from being well-understood and the uncertainties affect our understanding of polar ozone destruction, in particular in the northern stratosphere. However, the scientific field is very rapidly developing. This paper summarizes the current state of our knowledge of the microphysics and heterogeneous chemistry of polar stratospheric clouds with emphasis on liquid and solid particle thermodynamics and on kinetics of non-reactive gas uptake leading to particle growth. The consequences of the present uncertainties for the chemical processing of stratospheric air are briefly discussed. K e y w o r d s - Stratospheric aerosols; microphysics; heterogeneous chemistry INTRODUCTION Considerable progress has been made within the first half of this decade in our understanding of the composition and physical phase of polar stratospheric clouds (PSCs). Possibly most important is the recognition of the liquid state that PSCs can take on under certain conditions. Previously PSCs had been assumed to consist of solid particles only, like ice or nitric acid trihydrate (NAT), which are both crystalline. The three conventional types of PSCs distinguishable by Lidar (Light Detection And Ranging), type-la, type-lb and type-2, can tentatively be identified as NAT, ternary droplets and ice, respectively, although in-situ chemical analysis is still pending. Presently, open questions are how the solid particles form and how the chemical reactivity of the particles depends on their physical state. Furthermore, with highly resolved PSC observations from the Arctic becoming available there is evidence that not all PSCs may be explained in terms of liquid, NAT or ice particles under thermodynamic equilibrium conditions. Whether transient non-equilibrium states of the aerosol or other phenomena, like the currently discussed amorphous states, enable to understand these cases remains to be seen. THE PHYSICAL STATE OF PSCs Only one year after the discovery of the Antarctic ozone hole Crutzen and Arnold [1986] and Toon et al. [1986] suggested that PSCs would contain nitric acid. Especially the thermodynamically most stable nitric acid trihydrate (NAT = HNO3.3H20) could possibly account for the clouds existing at temperatures 5-7 K above the frost point. Based on this information the following 3-stage concept of PSC formation emerged [Poole and McCormick, 1988], which is used in most chemical transport models up to date. It was thought that: 1. at sufficiently low temperatures (T = 200-210 K) the ubiquitous stratospheric H2SO4/H20 aerosol particles could freeze and act as solid condensation nuclei;
280
Formation mechanisms of polar stratospheric clouds
281
2. around 195 K NAT formed on the frozen aerosol particles due to heterogeneous bimolecular nucleation and subsequent condensation of HNO3 and H20; 3. below 190 K ice can grow on NAT particles after heterogeneous unimolecular nucleation of H20. This 3-stage concept is summarized in Fig.1. Thermodynamic vapor pressures measured by Hanson and Mauersberger [1988] confirmed that NAT was indeed the thermodynamically most stable HNOa hydrate under stratospheric conditions. This also corroborated the 3-stage concept. Moreover, the concept sucessfully explained the PSC mes of the American Antarctic campaign (AAOE, 1987) which clearly showed the onset of particle growth when the saturation temperature with respect to NAT was reached [Fahey et al., 1989]. On the other hand, these Antarctic observations were made in late winter after air masses had been exposed to very cold conditions over extended periods of time and could not reveal how the PSCs were initially formed during early winter. However, the two main ideas after the confusion caused by the discovery of the ozone hole, heterogeoneous chemistry on PSCs [Solomon et al., 1986] and formation of stable HNOa hydrates, had been developed very quickly, and theory, laboratory experiment and field observation seemed to be in good agreement. CONVENTIONAL 3-STAGE PSC-MODEL The Different Types of Stratospheric Particles Below -195 K Below Frost Point (-187 K)
Initial Conditions o
O
O o
0
o
o
0
o
o
o
O
o
o
Sulfuric acid NAT = Type-I,
o
0
o
o
~)
o
NAT nucleation on the largest aerosol particles Nitric acid trihydrate
Ice = Type II particles,
o
0
o
o
N
o 0 o
"
Background H2SO 4 aerosol D
O
o0
l
Ice nucleation on the largest NAT particles; NAT is also incorporated into the ice
F ] Ice with a NAT impurity ~ Removal by sedimentation
Figure 1" Conventional 3-stage concept for PSC formation as it is currently applied in most of the large-scale chemical models. (1) Initial conditions with background H2SO4/H20 aerosol assumed to freeze at some temperature above 195 K. (2) Below about 195 K NAT nucleates directly from the vapor phase on the largest frozen aerosol particles. (3) Below about 187 K ice nucleates on NAT. Sufficiently large NAT and ice particles are removed due to sedimentation. Adapted from Drdla and Turco [1990]. At the turn of the decade several observations gave first hints that PSC particles do often not form according to the 3-stage concept. Lidar observations in the Arctic [Browell et al., 1990] indicated the existence of two distinct classes of type-1 PSCs, one of which (type-lb) consisting of a large number of non-depolarizing particles, which must be either spherical or, if non-spherical, have sizes much smaller than the Lidar wavelength [Toon et al., 1990]. Next, Hofmann et al. [1990] and Schlager et el. [1990] in two joint balloon experiments showed that in airmasses passing the Norwegian mountains PSCs formed only when supercooling with respect to NAT by 3-4 K was reached. Using a microphysical model Peter et al. [1992] established agreement between the 3-stage concept and these observations under two assumptions: first, the particles had undergone strong lee-wave-induced cooling briefly before observation; second, the crystalline compatibility
Peter
282
oc~s--- so2--- H~O4 OXIDATION+NUCLEATION
- 230
-220
UQUID AEROSOL ~i SULFURIC
-210
FROZEN?
I-- --I I O I i | I _
_
I _
.; NO
_,00
PSC-I I 190
NAT ~
~ H20/I"INO:~'42SO4
PSC-II ICE~H#:) t.,.-~../"e1420
Figure 2: Two possible pathways for the stratospheric aerosol upon cooling. Left path: conventional 3-stage concept ( / / / / S A T , \\\\ NAT, ]]11ice). Right path: aerosol remains liquid, takes up HNO3 and freezes out water ice below the frost point.
between NAT and the underlying frozen sulfuric acid had to be assumed very low. While the lee wave assumption is now fully accepted, a proper description of the microphysics in future must take account of liquid particles. Finally, also in 1990, referring to his laboratory measurements with Mauersberger, Hanson described the observation of HNO3/H20 liquid films prior to NAT formation [Hanson, 1990]. Stimulated by these observations Arnold [1992] suggested that H2SO4/H20 aerosols might remain liquid to temperatures close to the NAT equilibrium temperature, at which point it could take up HNO3. However, he had to rely on HNO3 solubility measurements of Reihs et aI. [1990] who themselves thought that HNOz would remain a minor component of the liquid aerosol. The low probability of the binary H2SO4/H20 aerosols to freeze as sulfuric acid tetrahydrate (SAT) was further corroborated by Luo et al. [1994a] who investigated the freezing process of binary H2SO4/H20 aerosols using classical homogeneous nucleation theory. These calculations depend crucially on thermodynamic input parameters, like the surface energy of the SAT embryo, which are still quite uncertain. Larsen [1994] showed Antarctic and Arctic denitrification could be explained within the 3-stage concept only if surface energies of the SAT embryo were much lower than assumed previously. On the other hand, based on observations in metallurgy MacKenzie et al. [1995] argued that the surface energy might actually be up to a factor of five larger than the one adopted by Luo et al. [1994a]. Hence, the question arose, wha~ would happen if temperatures fell below the NAT existence temperature and the particles were still liquid. Figure 2 summarizes the two possible pathways of the stratospheric aerosol with falling temperatures. The left path shows the conventional 3-stage concept: freezing of the H2SO4/H20 droplets, then sublimative NAT nucleation on the frozen particles, and finally water ice nucleation. The right path shows the droplets remaining liquid, take up H20 and HNO3, and eventually solidify below the frost point as water ice. It is at present not clear whether or not the surface layer of remaining acidic liquid will eventually freeze, possibly seeded by the ice nucleus. The acids might then form SAT and NAT layers on the ice which could possibly protect the ice core against evaporation. As mentioned above, NAT evaporates only 5-7 K above the frost point, and SAT even more than 25 K higher [Middlebrook et al., 1993].
283
Formation mechanisms o f polar stratospheric clouds
THERMODYNAMICS OF HNO3/H2SO4/H20 SOLUTIONS Vapor pressure measurements in the laboratory clearly indicate the onset of HNO3 uptake at low temperatures [Molina et al., 1993], but thermodynamic models are required to bridge the data gap when concentrations change from aqeous H2SO4 to aqeous HNO3. Ion-interaction models for the description of highly concentrated electrolytes have been used in many scientific disciplines. Using such thermodynamic models Carslaw et al. [1994], Tabazadeh et at. [1994] and Drdla el• al. [1994] successfully explained stratospheric particle measurements by Dye el, al. [1992] as liquid ternary solution droplets, justifying an earlier approach by Peter and Crutzen [1993]. Carslaw et al. used a form of the Pitzer ion-interaction model [Pitzer, 1987]. Ion-interaction models comprise electrostatic far field interactions between the ions similar to the theory of Debye and Hiickel [1923] and near field interactions due to two and three body collisions between individual ions or solvent molecules. The interaction parameters are determined from available thermodynamic data, and the model equations allow a thermodynamically self-consistent extrapolation to other conditions, for example lower temperatures. 10
.,...~
i
....
,~. .
E o
i ....
H2SO4
/ HNO
i
....
1 ....
60
i
1
i
I
I
!
1
i
.~
,---,,
3 / H20
i
1
1
i
l
""
o~ 40
~ ~ ....
1
r
E....j
20
. _O
)
E 0.1 _=
" ....
.~ r
O >
g l o .2
.. 0.01
2
I1) O
,,,
T9 i c e {
I
~ I
190
, I
~ ~
200 T[K]
J ~
I
~
,
~
l
0
(a)
TNAT
1 ~ J I ,v,,
1
,
210
10 -4• ' 185
190
195
200
T[K]
Figure 3: (a) Total particle volume observed as function of temperature onboard the American high-altitude research aircraft ER-2 on 24 January, 1989, during a flight northbound from Stavanger, Norway. Flight level was 55 mbar (about 19 km altitude). Amount of H2SO4 in the aerosol corresponds to normal background conditions (0.53 ppbv H2S04 upon evaporation). With decreasing temperatures the particles initially grow due to H20 uptake by the H2SO4/H20 binary (dotted line for 5 ppmv H20). Below 193 K volume increase is much stronger, and is in agreement with the combined H20 and HNO3 uptake by the ternary aerosol (heavy solid line for 10 ppbv HNO3, thin solid lines for 5 and 15 ppbv). NAT particle growth without nucleation barrier is shown by the dashed line. NAT and ice equilibrium temperatures are indicated by vertical arrows. (b) Liquid concentrations as functions of temperature (with 10 ppbv HNOa). Concentrations for H2SO4, HNO3 and HC1 in wt% for the ternary aerosol (solid lines) and for H2SO4 in the absence of HNO3 (dotted line). Note the change from linear to logarithmic scales below 2 wt%. Adapted from Carslaw et al. [1994]. Figure 3a shows the measurements of aerosol volume vs. temperature by Dye et al. [1992] on 24 January, 1989 north of Norway at about 55 mbar. The thick solid curve based on the thermodynamic model calculations of the coupled uptake of H20 and HNO3 by Carslaw et al. [1994] can fully explain the data. On the other hand it is clear that H20 uptake alone (dotted curve) cannot account for the observations. Also NAT formation, if one assumed thermodynamic equilibrium without a nucleation barrier, would lead to very different particle volumes with decreasing temperatures (dashed line). Figure 3b shows the strong changes in liquid composition of the aerosol particles leading to the observed volume growth in Fig.3a. A combination of decreasing temperatures and water uptake
284
Peter 200
~" 196 "~" 194
NAT
~a.192 E 190 188
186 30
40
50
60
70
80
90
1O0
Total Pressure (mbar)
Figure 4: Equilibrium temperatures for NAT (upper curve) and ice (frost point, lower curve) and temperature for the onset of substantial HN03 uptake into stratospheric ternary solution (STS) droplets. Adapted from MacKenzie et al., 1995. leads to an enhanced solubility of HNO3. Below about 193 K the coupled HNOa/H20 uptake leads to the transformation from a binary H2SO4/H20 solution above 195 K into a quasi-binary HNO3/H~O solution below 190 K. The temperature below which substantial amounts of HNO3 are taken up so that the gas phase starts to become depleted lies between the NAT equilibrium temperature and the frost point. This is illustrated in Fig.4 from MacKenzie et aI. [1995] as a function of pressure altitude. In parallel, effort was made to backup these results by experimental investigations of the freezing behaviour of solutions of stratospheric composition. Ohtake [1993] investigated only the H2SO4/H20 binary system and showed that the solutions tended to supercool considerably. Beyer et al. [1994] and Song [1994] investigated ternary solutions but did not exclude the possibility of freezing under stratospheric conditions. Finally, Koop et al. [1995] investigated various HNO3/H~SO4 mixtures with compositions as given by Fig.3b between 195 K and 188 K. They kept 1-ml samples of these solutions for hours at temperatures corresponding to the stratospheric values and thus excluded the possibility of homogeneous freezing above the frost point. In seeding experiments with compounds of stratospheric interest (magnetite representing meteoritic material, aluminium oxide representing rocket exhausts, silicates representing volcanic debris, etc.) they showed that these compounds did not induce heterogeneous nucleation. The overall conclusion was that under equilibrium conditions the stratospheric aerosol droplets are unlikely to freeze at temperatures above the water ice frost point. HNO3 HYDRATES OTHER THAN NAT With their mass-spectrometric laboratory investigations Hanson and Mauersberger [1988] characterized the thermodynamic properties (vapor pressures and coexistence curves with water ice and nitric acid monohydrate) of NAT. Based on FTIR-spectroscopic measurements Tolbert et al. [1992] suggested that NAT might occur in two modifications (a-NAT and ~-NAT), but whether this affects the thermodynamic properties of NAT is currently not clear. In any case NAT is the most stable nitric acid hydrate under stratospheric conditions. However, according to a general physical chemistry principle (Ostwald's rule) it is often not the stable hydrate which nucleates first, but rather some metastable phase which, only after some time, transforms into the stable end product. Worsnop et al. [1993] measured vapor pressures of the previously discovered metastable nitric acid dihydrate (NAD) which may exist close to the coexistence curve of NAT and the monohydrate (NAM). In the same work Worsnop and coworkers report a higher hydrate, possibly the decahydrate close to the NAT/ice coexistence curve. Two other hydrates of nitric acid have been identified which could play an important role during the initial formation of solid hydrates under
Formation mechanisms of polar stratospheric clouds
285
stratospheric conditions: Marti and Mauersberger [1994] presented evidence for the formation of HNO3.5H20, and Fox et al. [1995] identified the mixed hydrate H2SO4.HNO3.5H20. Whether these hydrates really form under stratospheric conditions is at present completely open, since observational techniques sophisticated enough to distinguish these hydrates in the stratosphere are not yet available. The thermodynamic properties of all other nitric acid hydrates, except NAT, are not well established. This is also true for NAD which has been the most extensively studied of these metastable phases. Worsnop et al. [1993] measured the H20 and HNO3 vapor pressures of NAD which indicated a maximum of the melting point curve of NAD around 220 K. On the other hand, Ji and Petit [1993] made a calorimetric measurements of the NAD melting points showing a more than 15 K higher melting point curve. At present it is not known what the reason for the strong disagreement between both measurements is. HYPOTHESES OF FROZEN PSC FORMATION Laboratory measurements in combination with the success of thermodynamic models in describing HNO3/H2SO4/H20 ternary solutions has lead to a reversal of our understanding of solid and liquid PSCs. Until three years ago PSCs were thought to consist of solid particles. Now we know that at least in the Arctic the clouds are often liquid: the appearence of type-lb clouds can be understood in terms of ternary droplets. However, the formation of solid particles (type-la) remains unclear. At present there are five hypotheses on stratospheric particle freezing: 1. the binary H2SO4/H20 solution droplets freeze out sulfuric acid tetrahydrate (SAT), then NAT nucleates sublimatively on SAT (the 3-stage concept); 2. the binary droplets initially remain liquid and take up large amounts of HNO3 when T < 193 K, and subsequently NAT freezes out from these ternary solutions [Molina et al., 1993]; 3. even the ternary droplets remain liquid a~ long as temperatures do not fall below Tic,, at which point water ice precipitates from the solutions [Koop et al., 1995]; 4. rapid temperature fluctuations may cause the formation of quasi-binary HNO3/H20 droplets with close to 1:3 stoichiometry, which freeze as NAT above the frostpoint [Meilinger et al., 1995]; 5. amorphous solid solutions of H2SO4 and H20 form at low temperatures and crystallize upon warming as sulfuric acid hemihexahydrate or SAT which may then serve as nuclei for amorphous solid HNO3/H20 particles (so-called type-lc PSCs) at temperatures below the NAT equilibrium point [Tabazadeh et al., 1995]. As described above, hypotheses (1) and (2) by now can be almost ruled out as the initial step of solid particle formation given the evidence from laboratory experiments. The 3-stage concept may still be an important mechanism in the Antarctic, where due to the low temperatures the particles remain solid for extended periods of time after their initial freezing. However, in the Arctic particles go through repeated cooling and warming cycles, which lead often to complete melting. And even in the Antarctic the perception that NAT does not readily nucleate is supported by a spectroscopic analysis of infrared field observations [Toon and Tolbert, 1995]. Freezing experiments under thermodynamic equilibrium conditions suggest that the a~rosol is unlikely to freeze above Tic~ and therefore support hypothesis 3. However, this effectively means that even in the Arctic stratosphere, where the winter remains warmer than in the Antarctic, temperatures are required to drop below the frost point. The present lack of knowledge on mesoscale meteorology makes it impossible to judge if mesoscale temperature fluctuations could satisfy this condition. Although this appears to be a strong requirement, wave structures in aerosol layers observed by airborne Lidar [Godin et al., 1994] indicate gravity-wave induced adiabatic cooling and warming cycles of 2.5 to 10 K.
286
Peter
Recently Meilinger et al. [1995] investigated the non-equilibrium changes in the type-lb droplet compositions due to rapid temperature fluctuations as they can occur downstream of mountain ridges (lee waves). They found that these fluctuations may cause the droplets to assume HNO3 concentrations much larger than under equilibrium assumptions, and these quasi-binary HNO3/H20 droplets possibly freeze as NAT. Further details on their calculations are given by Peter and Meilinger in the present Proceedings Volume. Finally, Tabazadeh et al. [1995] suggested that amorphous solid solutions of HNO3 and H20 (so-called type-lc PSCs) could form at a temperature inbetween the NAT equilibrium temperature ar.d the frost point. This could possibly explain observations of PSCs whose vapor pressures indicate neither ternary liquid nor NAT in equilibrium. Such HNO3/H20 particles evidently have to grow on solid cores, otherwise the HNO3 uptake would simply lead to ternary solutions. Tabazadeh et al. [1995] suggested that these cores could be SAT particles which themselves could result from crystallization of H2SO4/H20 particles in an amorphous, glassy state. Support for this hypothesis comes from laboratory experiments [Iraci et al., 1994] and field observations [Larsen et al., 1996] which seem to show a crystallization upon warming, a well-known process of glasses when heated above the glass point. On the other hand, the glass points of binary and ternary solutions of HNO3, H2SO4 and H20 are below 165 K, therefore the application of the concept of glasses to stratospheric aerosols appears to be questionable. In a very recent paper Ha et al. [1996] reveal that particular single component systems may develop a distinct solid phase some degrees above the glass point which is neither glass nor crystal. Whether such a polyamorphism does exist in the system of interest here is presently not clear. FORMATION O F I C E CLOUDS IN OROGRAPHICALLY INDUCED WAVES When temperatures fall a few degrees below the frost point water ice clouds (type-2 PSCs) develop either by nucleation from the vapor phase or by freezing from ternary solution droplets. The modeling of the full particle kinetics of ice and NAT particles according to the 3-stage concept, i.e. sublimative nucleation, diffusively limited growth, NAT core release by evaporating ice particles and SAT core release by evaporating NAT particles, is described by Toon et al. [1989], Poole et al. [1990], Drdla and Turco [1991], Peter et al. [1992] and Panegrossi et el. [1996]. The reliability of these modeling studies is limited mainly due to the neglect of the liquid phase and uncertainties in the nucleation processes. In-situ measurements of water ice clouds are rare, because the high-altitude research aircraft ER-2 deliberately avoids the turbulent regions around the ice clouds, and because stratospheric balloons are not easy to position in these mesoscale features. In the only in-situ type-2 PSC observations during the European ozone expedition EASOE particles were measured with a balloonborne optical particle counter over northern Sweden [Deshler et al., 1994]. The measurement was made on 27 January, 1992, when stratospheric temperatures above the Norwegian mountains had dropped below the frost point, causing water ice particles to nucleate. Deshler et al. [1994] attributed the observed strong temperature and particle density fluctuations to the lee wave perturbation above and behind the mountains. Individual particle layers with up to 5 #m large particles were visible. Using a meteorological mesoscale model with high-resolution orography [Volkert and Intes, 1992] calculated the small-scale temperature fluctuations exerted by the mountain lee waves and showed that lee-wave induced temperature changes can amount to many degrees Kelvin. Most surprising is the fact that the temperatures measured aboard the balloon are 5-10 K too high for ice to be in equilibrium. Pure water ice particles would evaporate within a few minutes under such subsaturated conditions. Peter et al. [1994] suggested that the HNO3 impurities in the ice might form small NAT clusters, which upon heating of the ice could accumulate at the ice
Formation mechanisms of polar stratospheric clouds Sedimentation- induced T > TNAT T < TNAT
287
NAT- coating
T < TICE T > TICE T > TNAT
o o-@--@--o
frozen
aerosol
H2SO4 ] H20
NAT grows on aerosol
T < Tic E
frozen a c q u i r e s aerosol NAT- H2SO4 / H20 coating
on NAT
Leewave- induced T > TIC E
ice crystal
ice grows
NAT- coating
T > TIC E T > TNA T T >> TNA T
o -O-o liquid aerosol
H2SO4 / H20 / HNO3
growing ice
shrinking ice
pressure
aerosol
crystalwith crystal acquires reduction liquid? NATNATdue to NAT- frozen? impurities coating rich layer
Figure 5: Upper panel: Sketch of the NAT coating mechanism proposed by Wofsy et al. (1990). Ice partilces sediment into warm regions where NAT particles have not yet formed and acquire a NAT coating which protects them against evaporation. Lower panel: Sketch of the evolution of an ice crystal with NAT-impurities nucleating in a liquid H2SO4/H20 aerosol particle. In this case ice evaporation is slowed due to a NAT-coating (Tice < T < TNAT) or due to a lowering of its vapour pressure (T > TNAT). surface since the water molecules evaporate faster than the NAT clusters. Eventually this could form a few NAT monolayers on top of the ice leading to a protection of the ice, i.e. to a slowing of the evaporation process due to a lowering of the vapor pressure (see Fig.5, upper panel). Wofsy et al. [1990] had suggested a similar coating mechanism for ice particles sedimenting into atmospheric layers which are too warm for pure ice to exist but still cold enough for NAT. The settling ice particles could then acquire a NAT coating on their way through these layers (Fig.5, lower panel). The increase of the ice particle lifetime due to NAT coating might strongly affect their importance as sites for heterogeneous chemical reactions in the lower Arctic stratosphere. CHLORINE ACTIVATION ON LIQUID AND THE SOLID PARTICLES Our understanding, not only of the thermodynamics, but also the reaction kinetics of liquid stratospheric particles has advanced considerably within the last three years. The accurate determination of (non-reactive) solubilities of individual reaction partners in the HNO3/H2SO4/H20 solutions is also central for a correct description of the reactivity. Figure 6 shows measurements by Elrod et al. [1995] and Hanson and Ravishankara [1994] of the reaction probability 7 of CION02 for the reaction C1ON02 + HC1 pmide ~C12 + HN03 (1) in H2SO4/H~O binary and HNO3/H2SO4/H20 ternary solutions along with data for NAT (dotted line) from Hanson and Ravishankara [1993]. In contrast to reaction probabilities on solids there is a reliable framework for theoretical treatment of liquid phase reactions [Hanson et al., 1994]. If reactant concentrations are not too high the reaction probability can be calculated from the solubility and diffusion of the two reactants into the liquid and can be expressed as "7 -~ 4RT/~clONO~ (DIgII[Hc1])z/2/vcloNo2.
(2)
288
Peter ~w
>,
i
|
,
l
w
i
l
!
0.12
.m
o
I,,.,..
c~
0.08
tO
0.04 ~176176176
0 194
1
I 196
I
[]
~'~
! 198
i
200
202
T/K
Figure 6: Reaction probabilities for HCl+C1ONO2 in liquid aerosols (solid and dashed lines) and on NAT (dotted line) as a function of temperature assuming 5 ppmv H20, 0.5 ppbv H2SO4 and 2 ppbv HC1 at 100 mbar. Solid line: calculated from Eq.(2), see text. Dashed line: same as solid line, but assuming in addition 10 ppbv HN03 to be present. Open squares: measurements with HNOa [Elrod et al., 1995]. Filled squares: measurements without HNO3 [Elrod et al., 1995]. Filled triangles: without HNOa estimated from Hanson and Ravishankara [1994].
Figure 6 shows the measured reaction probabilities in comparison with theoretical curves (solid line: H2SO4/H20; dashed line: H2SO4/HNOz/H20) calculated from Eq.(2). The solubility of HC1 is calculated using the model vapor pressures given by Luo et al. [1995]. The liquid diffusion constant (Dr) is calculated from the cubic cell model [cp. Luo et al., 1994b]. The second-order rate coefficient (K zx) is calculated assuming the reaction to be diffusion limited, and the Henry's law constant of chlorine nitrate is estimated as H*aoso2 -- 103 mol 1-1atm -1 at 202 K and assuming - 6 kcal mo1-1 for the enthalpy of solvation [Hanson and Ravishankara, 1994]. The good agreement between the measurements and the theoretical estimate suggests that the temperature dependence of the reaction probability is governed by the solubility of HC1 in the liquid aerosol. Furthermore, it supports the suggestion by Hanson and Ravishankara that C1ON02 probably remains undissociated in the liquid, so that its interaction with the liquid involves mainly hydrogen bonding. At low temperatures chlorine activation can occur efficiently not only in binary and ternary solution droplets, but also on the surfaces of solid stratospheric particles. In Fig.6 the dotted curve shows the reaction probability for Eq.(1) on NAT [Hanson and Ravishankara, 1993]. Abbatt and Molina [1992] investigated the same reaction, but for higher HC1 concentrations. A crude extrapolation of their data to 2 x 10-r mbar HC1 shows that the corresponding 7 is more than one order of magnitude lower than the dotted line in Fig.6. There still appears to be a significant uncertainty in the solid phase reaction probabilities. In any case Fig.6 shows that the 7's on the droplets can be higher than those on solid NAT particles, as also pointed out by Ravishankara and Hanson [1996]. CONCLUSIONS The thermodynamics of stratospheric solution droplets have been extensively studied in the last three years. The thermodynamics of nitric acid trihydrate (HNOa.3H20), a major candidate expected to form PSCs, is well-established. There is some evidence that other hydrates of nitric acid could also play a role in the PSC formation mechanism, like HNOa.2H20, HNOa.bH20,
Formation mechanisms of polar stratospheric clouds
289
HNO3.10H20 or H2SO4.HNO3"5H20, but knowledge of the thermodynamics of these particles is still far from complete. Our understanding of the nucleation and kinetics of non-reactive gas uptake by solid and liquid particles is currently developed to different degrees. The growth kinetics of solid stratospheric particles have been investigated since the late eighties. Main uncertainties concern the following nucleation processes: freezing of the H~SO4/HNO3/H20 aerosol particles in the form of nitric or sulfuric acid hydrates, formation of nitric acid hydrates (NAT, NAD, etc.) on frozen H2SO4 surfaces, and nucleation of water ice on these hydrates. Also, once nucleated, it is at present uncertain whether ice particles grow as pure water ice or whether HNO3 may co-condense, possibly leading to the formation of protective coatings. To these uncertainties in the microphysis of PSCs add the uncertainties concerning the chemical reactivity of chlorine gases on liquid and solid surfaces. As a consequence, our understanding of PSC formation and chlorine processing in the Arctic winter stratosphere is incomplete. This limits our diagnostic and, in particular, our prognostic capabilities. Most urgent is a physico-chemical insitu analysis of PSC particles together with Lagrangian measurements along a complete life-cycle of the PSCs. Acknowledgements and notes: The author is indebted to Drs. Beiping Luo, Kenneth Carslaw and Thomas Koop for their scientific assistance in the preparation and to Ms Jennifer Carslaw for carefully reading the manuscript. Special thanks also to Ms Iris Bambach and Ms Gerhild Feyerherd of the MPI drawing office. This paper is a part of a review to be published in Solar System Ices (eds. M. Festou, C. de Bergh), in: Astrophysics and Space Science Library Series, Kluwer Acad. Publ., 1996. REFERENCES Arnold, F., Stratospheric aerosol increases and ozone destruction: Implications from mass spectrometer measurements, Ber. Bunsenges. Phys. Chem., 96, 339-350, 1992. Abbatt, J.P.D., Molina, M.J., The heterogeneous reactions of HOCI+HC1 ~ C12+H20 on ice and nitric acid trihydrate: Reaction probabilities and stratospheric implications, Geophys. Res. Lett., 19, 461-464, 1992. Beyer, K.D., Seago, S.W., Chang, H.Y., Molina, M.J., Composition and freezing of aqueous H2SO4/HNO3 solutions under polar stratospheric conditions, Geophys. Res. Lett., 216, 871-874, 1994. Browell, E.V., Butler, C.F., Ismail, S., Robinette, P.A., Carter, A.F., Higdon, N.S., Toon, O.B., Schoebed, M.R., Tuck, A.F., Airborne Lidar observations in the wintertime Arctic stratosphere: Polar stratospheric clouds, Geophys. Res. Left., 17, 385-388, 1990. Carslaw, K.S., Luo, B.P., Clegg, S.L., Peter, Th., Brimblecombe, P., Crutzen, P.J., Stratopsheric aerosol growth and HNO3 gas phase depletion from coupled HNO3 and water uptake by liquid particles, Geophys. Res. LetL, 21, 2479-2482, 1994. Crutzen, P.J., Arnold, F., Nitric acid cloud formation in the cold Antarctic stratosphere: A major cause for the springtime "ozone hole", Nature, 32~, 651-655, 1986. Debye, P., Htickel, E., Zur Theorie der Elektrolyte. I. Gefrierpunktserniedrigung und verwandte Erscheinungen, Phys. Z., 24, 185-206, 1923. Deshler, T., Peter, Th., Mtiller, R., Crutzen, P.`I., The lifetime of leewave-induced ice particles in the Arctic stratopshere: I. Balloonborne observations, Geophys. Res. I,ett., 21, 1327-1330, 1994. Drdla, K., Turco, R.P., A one-dimensional model of Type-I and Type-II PSC formation with temperature oscillations, XV. General Assembly EGS, Copenhagen, 1990. Drdla, K., Turco, R.P., Denitrification through PSC formation: A 1-D model incorporating temperature oscillations, J. Atmos. Chem., 12, 319-366, 1991. Drdla, K., Tabazadeh, A., Turco, R.P., Jacobson, M.Z., Dye, ,I.E., Twohy, C., Baumgardner, D., Kelly, K.K., Chan, R.P., Loewenstein, M., Analysis of the physical state of one Arctic polar stratospheric cloud based of observations, Geophys. Res. I,ett., 21, 2473-2478, 1994.
290
Peter
Molina, M.J., Zhang, R., Wooldridge, P.J., McMahon, J.R., Kim, J.E., Chang, H.Y., Beyer, K.D., Physical chemistry of the H2SO4/HNO3/H20 system: Implications for polar stratospheric clouds, Science, 26i, 1418-1423, 1993. Ohtake, T., Freezing points of H2SO4 aqueous solutions and formation of polar stratospheric clouds, Tellus, 45B, 138-144, 1993. Panegrossi, G, Fua, A., Fiocco, G., A 1-D model of the formation and evolution of polar stratospheric clouds, d. Atmos. Chem., 23, 5-35, 1996. Peter, Th., Crutzen, P.J., The role of stratospheric cloud particles in polar ozone depletion - An overview, J. Aerosol Sci., 2~, Sl19-S120, 1993. Peter, Th., Mfiller, R., Drdla, K., Petzoldt, K., Reimer, E., A micro-physical box model for EASOE: Preliminary results for the January/February 1990 PSC event over Kiruna, Bet. Bunsenges. Phys. Chem., 96, 362-367, 1992. Peter, Th., Mfiller, R., Crutzen, P.J., Deshler, T., The lifetime of leewave-induced ice particles in the Arctic stratosphere: II. Stabilization due to NAT-coating, Geophys. Res. Lett., 21, 1331-1334, 1994. Pitzer, K.S., A thermodynamic model for aqueous solutions of liquid-like density, Reviews Minerology, I7, 97-142, 1987. Poole, L.R., McCormick, M.P., Polar stratospheric clouds and the Antarctic ozone hole, J. Geophys. Res., 93, 8423-8430, 1988. Poole, L.R., Pitts, M.C., Polar stratospheric cloud climatology based on Stratospheric Aerosol Measurement II observations from 1978-1989, J. Geophys. Res., 99, 13,083-13,089, 1994. Poole, L.R., Solomon, S., Gandrud, B.W., Powell, K.A., Dye, J.E., The polar stratospheric cloud event of January 24, 1989 Part 1. Microphysics, Geophys. Res. Lett., 17, 537-540, 1990. Ravishankara, A.R., Hanson, D.R., Differences in the reactivity of type-I polar stratospheric clouds depending on their phase, J. Geophys. Rex., 101, 3885-3890, 1996. Reihs, C.M., Golden, D.M., Tolbert, M.A., Nitric acid uptake by sulfuric acid solutions under stratospheric conditions: determination of Henry's law solubility, J. Geophys. Res., 95, 16545-16550, 1990. Schlager, H., Arnold, F., Hofmann, D., Deshler, T., Balloon observations of nitric acid aerosol formation in the Arctic stratosphere: I. Gaseous nitric acid, Geophys. Res. Lett., 17, 1275-1278, 1990. Solomon, S., Garcia, R.R., Rowland, F.S., Wuebbles, D.J., On the depletion of Antarctic ozone, Nature, 321, 755-758 , 1986. Song, N., Freezing temperatures of H2SO4/HNO3/H20 mixtures: Implications for polar stratospheric clouds, Geophys. Res. Lett., 21, 2709-2712, 1994. Tabazadeh, A., Turco, R.P., Drdla, K., Jacobson, M.Z., A study of Type I polar stratospheric cloud formation, Geophys. Res. Lett., 21, 1619-1622, 1994. Tabazadeh, A., Toon, O.B., Hamill, P., Freezing behavior of stratospheric sulfate aerosols inferred from trajectory studies, Geophys. Res. Lett., 22, 1725-1728, 1995. Tolbert, M.A., Koehler, B.G., Middlebrook, A.M., Spectroscopic studies of model polar stratospheric cloud films, Spectrochim. Acta, 48A, 1303-1313, 1992. Toon, O.B., Hamill, P., Turco, R.P., Pinto, J., Condensation of HNO3 and HC1 in the winter polar stratospheres, Geophys. Res. Lett., 13, 1284-1287, 1986. Toon, O.B., Turco, R..P., Jordan, J., Goodman, J., Ferry, G., Physical processes in polar stratospheric ice clouds, J. Geophys. Res., 94, 11,359-11,380, 1989. Toon, O.B., Browell, E.V., Kinne, S., Jordan, J., An analysis of Lidar observations of polar stratospheric clouds, Geophys. Res. Lett., 17, 393-396, 1990. Toon, O.B., Tolbert, M.A., Spectroscopic evidence against nitric acid trihydrate in polar stratospheric clouds, Nature, 375, 218-221, 1995. Volkert, H., Intes, D., Orographically forced stratospheric waves over northern Scandinavia, Geophys. Res. Lett., 19, 1205-1208, 1992. Wofsy, S.C., Salawitch, R..J., Yatteau, J.H., McElroy, M.B., Gandrud, B.W., Dye, J.E., Baumgardner, D., Condensation of HNOs on falling ice particles: Mechanism for denitrification of the polar stratosphere, Geophys. Res. Lett., 17, 449-452, 1990. Worsnop, D.R., Fox, L.E., Zahniser, M.S., Wofsy, S.C., Vapor pressures of solid hydrates of nitric acid: Implications for polar stratospheric clouds, Science, 259, 71-74, 1993.
Formation mechanisms of polar stratospheric clouds
291
Dye, J.E., Baumgardner, D., Gandrud, B.W., Kawa, S.R., Kelly, K.K., Loewenstein, M., Ferry, G.V., Chan, K.R., Gary, B.L., Particle size distributions in Arctic polar stratospheric clouds, growth and freezing of sulfuric acid droplets, and implications for cloud formation, J. Geophys. Res., 97, 8015-8034, 1992. Elrod, M.J., Koch, R.E., Kim, J.E., Molina, M.J., HCI vapour pressures and reaction probabilities for C1ONO2+HC1 on liquid H2SO4-HNO3-HCI-H20 solutions, Faraday Discuss., 100, in press, 1995. Fahey, D.W., Kelly K.K, Ferry G.V., Poole L.R., Wilson J.C., Murphy, D.M., Loewenstein, M., Chan K.R., In-situ measurements of total reactive nitrogen, total water, and aerosol in a polar stratospheric cloud in the Antactic, J. Geophys. Res., 94, 11299-11315, 1989. Fox, L.E., Worsnop, D.R., Zahniser, M.S., Wofsy, S.C., Metastable phases in polar stratospheric aerosols, Science, 267, 351-355, 1995. Ha, A., Cohen, I., Zhao, X., Lee, M., Kivelson, D., Supercooled liquids and polyamorphism, J. Phys. Chem., 100, 1-4, 1996. Hanson, D.R., The vapor pressures of supercooled HNO3/H20 solutions, Geophys. Res. Lett., 17, 421-423, 1990. Hanson, D., Mauersberger, K., Laboratory studies of the nitric acid trihydrate: Implications for the south polar stratosphere, Geophys. Res. Lett., 15, 855-858, 1988. Hanson, D.R., Ravishankara, A.R., Reaction of C1ONO2 with HCI on NAT, NAD, and frozen sulfuric acid and hydrolysis of N205 and C1ONO2 on frozen sulphuric acid, J. Geophys. Res., 98, 22,931-22,936, 1993. Hanson, D.R., Ravishankara, A.R., Reactive uptake of C1ONO2 onto sulfuric acid due to reaction with HC1 and H20, J. Phys. Chem., 98, 5728-5735, 1994. Hanson D.R., Ravishankara A.R., Solomon S., Heterogeneous reactions in sulfuric acid aerosols: A framework for model calculations, J. Geophys. Res., 99, 3615-3629, 1994. Hofmann, D.J., Deshler, T., Arnold, F., Schlager, H., Balloon observations of nitric acid aerosol formation in the Arctic stratosphere: II. aerosol, Geophys. Res. Lett., 17, 1279-1282, 1990. Iraci, L.T., Middlebrook, A.M., Wilson, M.A., Tolbert, M.A., Growth of nitric acid hydrates on thin sulfuric acid films, Geophys. Res. Lett., 21, 867-870, 1994. Ji, K., Petit, J-C., Identication par microcalorimetrie d'un nouvel hydrate de l'acide nitrique pouvant jouer un role dans la chimie heterogene stratospherique, C. R. Acad. Sci., 316, 1743-1748, 1993. Koop, T., Biermann, U.M., Raber, W., Luo, B.P., Crutzen, P.J., Peter, Th., Do stratospheric aerosol droplets freeze above the ice frost point?, Geophys. Res. Lett., 22, 917-920, 1995. Larsen, N., The impact of freezing of sulfate aerosols on the formation of polar stratospheric clouds, Geophys. Res. Lett., 21, 425-428, 1994. Larsen, N., Knudsen, B., Rosen, J.M., Kjome, N.T., BMloonborne backscatter observations of type 1 PSC formation: Inference about physical state from trajectory analysis, Geophys. Res. Lett., 23, 1996. Luo, B.P., Peter, Th., Crutzen, P., Freezing of stratospheric aerosol droplets, Geophys. Res. Lett., 21, 1447-1450, 1994a. Luo, B.P., Clegg, S.L., Peter, Th., M/iller, R., Crutzen, P.J., HCL solubility and liquid diffusion in aqueous sulfuric acid under stratospheric conditions, Geophys. Res. Lett., 21, 49-52, 1994b. Luo, B.P., Carslaw, K.S., Peter, TH., Clegg, S., Vapour pressures of H2SO4/HNO3/HC1/ HBr/H20 solutions to low stratospheric temperatures, Geophys. Res. Lett., 22, 247-250, 1995. MacKenzie A.R., Kulmala, M., Laaksonen, A., Vesala, T., n the formation of type 1 polar stratospheric cloud particles, Geophys. Res. Lett., 21, 1423-1426, 1994. Marti, J.J., Mauersberger, K., Evidence for nitric acid pentahydrate formed under stratospheric conditions, J. Phys. Chem., 98, 6897-6899, 1994. Meilinger, S., Koop, T., Luo, B.P., Huthwelker, Th., Carslaw, K.S., Krieger, U., Crutzen, P.J., Peter, Th., Size-dependent Stratospheric Droplet Composition in Mesoscale Temperature Fluctuations and their Potential Role in PSC Freezing, Geophys. Res. Lett., 22, 3031-3034, 1995. Middlebrook, A.M., Iraci, L.T., McNeill, L.S., Koehler, B.G., Wilson, M.A., Saastad, O.W., Tolbert, M.A., Fourier transform-infrared studies of thin H2SO4/H20 films: Formation, water uptake, and solid-liquid phase changes, J. Geophys. Res., 98, 20,473-20,481, 1993.
AEROSOL PRODUCTION CAUSED BY CIVIL AIR TRAFFIC: AN OVERVIEW OF NEAR-FIELD INTERACTIONS B. I(,;~RCHER AND B.P. LUO Universitit Miinchen, Hohenbachernstr. 22, D-85354 Freising, Germany
A b s t r a c t - This contribution provides an overview of the physico-chemical processes which generate and modify particles in the plumes of cruising jet aircraft. The primary aerosol types are characterized and their complex interaction pathwavs are investigated. Open questions and further research issues are discussed. K e y w o r d s - Aircraft emissions; Binary nucleation; Heteromolecular condensation; Combustion aerosols (soot)" Freezing nucleation; Visible contrail formation AEROSOL DYNAMICS
-
AN OVERVIEW
Because of the rapid growth of commercial aviation, there is an increasing concern about the environmental effects of jet aircraft emissions on the atmosphere (WMO 1995). Particulate emissions form airliners cruising in the upper troposphere and lower stratosphere can potentially increase the mass and chemical reactivity of the global background aerosol layer, thereby affecting both chemical properties of the atmosphere and the radiative energy balance and, hence, global climate. heteromolecular gas-to-particle homogeneous
conversion J
nuclea.~
primary
aerosol
volatile .., droplets "
ael sol
~
soot particles
2 x 1011 cm-2s -1 (per unit ~ 10'a area of soot per unit time) for the freezing rate, using ,2 rs = 20nm. The solid line depicts maximum values 3}rz 10 ""- (0.3,60) for an ideal nucleus surrounded by a pure water coating. 10" 215 220 225 230 235 The freezing rates are rapid enough for 0 < 60 ~ in solutions containing less than 20wt-% H2SO4. The freezing temperature (K) point depression is too pronounced for higher acidities; Fig.4. Heterogeneous freezing rates verimperfect matching between the crystal lattice of the ice sus temperature. Legends indicate (W, 8). germ and the soot surface hampers nucleation for larger 8-values. We finally note that in subsonic jet plumes, homogeneous freezing can compete with heterogeneous freezing only at low (< 215 K) ambient temperatures. Clearly, it is essential to obtain more experimental information on gas-to-liquid and liquid-to-solid phase transitions involving H20 and H2SO4 triggered by soot in order to assess open aircraft-related problems concerning contrail-cirrus interaction, anthropogenic cloud formation, and impact on atmospheric chemistry. REFERENCES Abraham, F.F. (1974) Homogeneous Nucleation Theory. Academic Press, New York, p.99 Hofmann, D.J., Rosen J.M. (1978) Geophys. Res. Lett. 5, 511 Fahey, D.W. et al. (1995) Science 270, 70 K&rcher, B., Peter, Th., Ottmann. R. (1995) Geophys. Res. Lett. 22. 1501 Ki~rcher, B. et al. (1996a) J. Atmos. Sci., in press K'~rcher, B., Hirschberg, M.M., Fabian, P. (1996b) J. Geophys. Res.. in press K~cher, B. et al. (1996c) Manuscript in preparation Kuhnala M., Laaksonen, A. (1990) J. Chem. Phys. 93, 696 World Meteorological Organization (1995) Report No.37, WMO, Geneva, Switzerland
PARTICLE
FORMATION IN JET AIRCRAFT EXHAUSTS AND CONTRAILS FOR DIFFERENT SULFUR CONTAINING FUELS U. S C H U M A N N
DLR, Institut ftir Physik der Atmosph~ire, Oberpfaffenhofen, D-82230 Wessling, Germany Abstract. - A series of experiments has been performed observing contrail formation of twin-engine jet aircraft (ATTAS-VFW 614 and Airbus A310-300) run with different sulfur containing fuels on the two engines at the same time. The fuel sulfur mass content was varied from 2 to 5500 ppm. The results suggest that contrail particles form mainly from soot particles. The higher the sulfur content the more of the soot particles get activated as condensation nuclei. Particles start condensing in the liquid phase but have to freeze quickly, and the water vapor mass accommodation coefficient must be larger than about 0.2 both for liquid and ice particles, in order to form a visible contrail within 25 m after the aircraft as observed. Keywords - liquid and ice particles, visibility, accommodation coefficient, sulfuric acid INTRODUCTION Contrails (condensation trails) from engine exhaust of high-flying aircraft may influence the climatological and chemical state of the atmosphere. It is commonly assumed (for review see Schumann, 1996), that contrails form when isobaric mixing between the hot and humid exhaust gases and cold ambient air leads to a mixture reaching sufficient saturation so that many liquid or ice particles form which grow to a size such that the optical thickness becomes large enough to form a visible line cloud. Thermodynamic theories (e.g., Appleman, 1953) predict that contrails form when the humidity in the plume reaches liquid saturation so that droplets form on preexisting cloud condensation nuclei (CCN). This requires that the ambient temperature is below a threshold temperature o f - 4 0 to -60~ depending on altitude and ambient humidity. The nature of the required CCN and the phase of the water particles forming is open. The phase of the water particles decides on the critical temperature of contrail formation, their length and the distance of onset behind the aircraft. The required CCN may come either from ambient aerosol mixed into the exhaust plume, or from homogeneously nucleated sulfuric acid-water droplets, or from soot which has been activated by heterogeneously condensing sulfuric acid with water. Measured aerosol number densities in young contrails are much larger than in ambient air (Pitchford et al., 1991; Schumann et al., 1996), and model estimates (K~ircher et al., 1996) show that the amount of aerosol available in ambient air is too small to explain the formation of visible contrails. Also homogeneously formed sulfuric-acid water droplets in jet engine exhaust plumes are too small to become frozen and to grow to a size where they may become visible within reasonable distance behind the aircraft (K~ircher et al., 1995). In order to identify the impact of fuel sulfur content, a series of experiments has been performed in which twin-engine jet aircraft were operated with different sulfur containing fuels on the two engines at the same time. The aircraft was either a mid-size jet of type VFW 614, known as the ATTAS test aircraft of DLR, or an Airbus A310-300. For fuel sulfur contents of 2 and 250 ppm (Busen and Schumann, 1995), and flight conditions close to the critical conditions for liquid phase contrail formation, no visible differences were observed, Both contrails formed about 25 to 30 m behind the engines, at a plume age of about 0.2 s, under ambient conditions for which the exhaust plume reached a maximum relative humidity RH during mixing below 103%, according to homogeneous thermodynamic mixing theory. As noted in K~ircher et al. (1996), this maximum saturation appears to be too small to allow for formation of a visible contrail without freezing. In order to form a visible contrail, the number of particles Npart per unit volume, the particle radius r and the geometrical thickness of the contrails D must be large enough to form a cloud with optical thickness Xop,- rtr2DQ~xtNpar,larger than at least 0.02. In other experiments with the ATTAS, for 170 and 5500 ppm (Schumann et al., 1996) and with the A310-300 for about 220 and 3000 ppm sulfur content (preliminary), high sulfur causes contrails to form about 5 m earlier than for low sulfur plumes. Particle measurements (Schumann et al., 1996) 296
Particle f o r m a t i o n in j e t aircraft exhausts a n d contrails
297
indicate that the number of particles larger than 7 nm measured in the plume originated mainly from emitted soot particles. The number of particles in this size range increases by less than 50 % when the sulfur content is increased by a factor of 30. Color differences were observed between contrails formed from two engines burning fuels with different sulfur content (Schumann et al., 1996) and this color difference can be explained with a larger fraction of activated soot particles for high sulfur contents (Gierens and Schumann, 1996). The present paper mainly discusses the process which leads to a visible contrail within 0.2 s plume age for the first experimental case with rather low sulfur contents. PARAMETERS AFFECTING CONTRAIL FORMATION A fraction rl of the combustion heat, corresponding to the propulsion efficiency of the aircraft, is used to propel the aircraft and is converted to kinetic energy of turbulence and vortex flows in the aircraft wake. This fraction is not available to heat the young exhaust plumes in which the contrail forms. For this reason, the contrail formation has to be computed as if only 1 - q of the combustion heat Q is released into the exhaust plume (Busen and Schumann, 1995). Moreover, modem jet engines consist of a core engine and a bypass. The core exhaust contains all the water vapor resulting from the combustion process but only part of the heat. The other fraction of heat leaves the engine with the bypass air. This inhomogeneous heat release causes higher maximum values of relative humidity in the core exhaust plume (K~ircher, 1994). A large-eddy simulation of the exhaust jets from the core and bypass engine (Schumann et al., 1995) has shown that it takes an order of 0.1 s to mix the bypass and the core exhaust homogeneously and the local relative humidity may be about 10 to 20% larger than for the homogeneous exhaust case. In this paper, we present results of a simple particle growth model to identify the conditions under which the visibility of the observed contrail could be explained. The computations are performed for values of the aircraft and atmospheric parameters as given in Busen and Schumann (1995): Aircraft speed V= 115 m s-l, fuel flow rate rhF= 125 g s-l, specific combustion heat Q = 4 3 MJ kg -~, water vapor emission index EIH2o= 1.21, propulsion efficiency 1"1= 0.14, ambient pressure p = 30230 Pa, ambient temperature TE = 222.95 K, ambient relative (liquid) humidity RHE = 45%. Within the range of uncertainty, the given temperature and humidity values are those which result in the highest possible relative humidity in the homogeneous exhaust of up to 103%. The emission index of soot is taken as Elsoot= 0.5 g kg -1 (Schumann et al., 1996). We consider various alternatives. First, we vary the the particle formation processes: 1) The particles form by liquid condensation on CCN and stay liquid while growing; 2) the particles form when the saturation is large enough to form liquid particles which then freeze immediately and grow as ice particles; 3) the particles form as ice particles when the relative humidity exceeds the saturation humidity of ice. In all three cases we assume that the CCN are soot particles, possibly activated by sulfuric acid. We assume that all soot particles have the same radius rsoot (which is a free parameter of order 30 nm), all soot particles form CCN, and the number of soot particles corresponds to an emission index EIp~ = Elsoo,/(Psoot4rtr3ooJ3), where P~oo~= 2 g cm -3 is the soot density. Second, we consider exhaust gases from the engine core and bypass engines to be either 1) well mixed from the beginning (b = 0) or 2) separated so that a fraction b = 0.2 of the exhaust heat is not included in the contrail air initially but becomes mixed in slowly, causing higher maximum saturation (at most 107%). Third, we vary the mixing rates of the plume exhaust gases with the ambient air: 1) we assume linear increase of the plume cross-section with time from a plume diameter of Do = 1 m to Dl = 4 m. This value is larger than the observed contrail diameters of about 2.2 m (Schumann et al., 1996). 2) We assume that the plume exhaust mixes suddenly with ambient air to Do = D~ = 3.2 m, with maximum relative humidity, and then stays at constant diameter without further growth. Finally, we vary the water vapor mass accommodation coefficient Otc between 0.01 and 1 (in steps of 0.01) to determine the minimum value of Ctc required to form a visible contrail. THE MODEL The mass specific enthalpy h and the mass specific total water content m in the plume are h = hE + Q(1 - rl)(1 - b)/N, m = mE + EIH,o/N. Here N is the dilution factor, i.e., the mass of plume air
298
Schumann
per burned fuel mass, N = pAV/rhF, for given air density p = p](R~rT), and plume cross-section A = Ao + (A~-Ao)t/t~, A = reD2~4. The bypass heat fraction b is computed from db/dt = - b/xh with % = 0.1 s, and initial values of b either 0 or 0.2. The total water is composed of water vapor my and condensed water (liquid or ice) me, m = rn~ + mc. The temperature T in the exhaust follows from the enthalpy and the release of latent heat L, h - hE = c r ( T - T E ) - Lmc. The amount of condensed water is a function of the particle radius r and number N0art per volume, mc = Npa.(4/3)rcr3pJp~. Here Np.rt = EIpar,p J N . The droplet size increase is computed from Pruppacher and Klett (1980), dr r d--T=
S- Y
+
pcRv T esat(T)D'v
-1 ~
The saturation is S = (mvRH2oP)/(Rairesat(T)), where esa, is either the saturation pressure for liquid water or ice depending on the assumed particle phase, and Y accounts for the Kelvin effect Y= (2~)/(RH,oTpcr) due to surface stresses cy between ice or liquid and vapor. In the present computations we do not account for the Raoult effect which is small for the considered small sulfur contents. The effective heat conduction coefficient and effective diffusivity for water vapor are Dv
Kair
K' =
1/2
r r + )~ +
~
,
D'
v --
1/2 '
r r + )~ + ~
RairT
(2)
1
RH2oT
for given molecular conductivity Kant, diffusivity Dr, and mean free path of air molecules 2,,. We set a r = 0.7, and vary O~c. All the material parameters are taken from Pruppacher and Klett (1980). The extinction efficiency Q~xt of the particles is computed using Mie theory, assuming optical diffraction indices of (1.333,-1.69x10 -9) for liquid water and (1.311,-3.11x10 -9) for ice, a wavelength of 550 nm for solar light, and for a geometrical contrail thickness of D = 1 m. Eq. (1) is integrated with an Euler scheme in time steps of 2 ms, starting from r = r~oo, with the constraint r >_ r~oot. The integrations are performed from time t = 0 until t = fi = 0.2 s corresponding to the observed contrail. Figure 1 shows a typical result. IE-I
300
d
200 /
/,
150
pc /t
,,
a
I d
IO0
r
,,
/
,,
,
/
/
5
/
r
/
1E-5-
/ ,,
.II0 TIME IN S
.I15
,,1 ,' /
/
1E-4-
/
I
.05
/
~
s
IE-3,
/
/
/" 50
IE-2 5
/
/
r
ss S~ 9
/
//
4
that hit
occurances.
258 554 367 179
Fletcher,
N.H.
( 1966
)
Fletcher,
N.H.
( 1968
) Phil.
( 1988
) 12
in
a
253 529 352 174
without
field
molecular
in
this
It
is
area
action.
complex are,
assumed
growth
generally,
that
these
The th
Physics Mag. ICNAA.
of 18,
Rainclouds, 1287
.
in sta-
occurran-
REFERENCES
V.V.
func-
Bo ( w i t h -
coincide.
Klingo,
a
T,K
B B+ BO B_
and
435
sign
layer-
T~
area
Su/=
ion
surface
GE/kT
1
at
ice
I
jl>
1460
Cambridge.
DIRECTED ICE NUCLEI MODIFICATION BY VARIATION OF AEROSOL PARTICLES COMPOSITION A.D.MALKINA, V.V.PATRIKEEV, N.S.KIM, A.V.SHKODKIN Moscow State Agro-Engineering University,Department of Physics, Timiriazevskaya Street 58, Moscow, 127550, Russia Institute of Organic Chemistry, Russian Academy of Science,Leninski pr., 47, Moscow, 117334, Russia Institute of Experimental Meteorology, pr.Lenina, 82, Obninsk, 249020, Russia
Abstract- The paper is devoted to copper acetylacetonate as a new seeding agent. This nucleating substance meets requirements to cloud seeding materials: the output of active nuclei per gram of the substance is of the order of 10 ~4, particles at - 10~ its threshold temperature is about - 2~ Artificial ice nuclei can be easily produced by evaporation and condensation, their ice nucleation ability is not deactivated by ultraviolet radiation. This seeding is an inexpensive material and environmentally useful. Create "express"- method for evaluating the nucleative properties a greate variety of substances. Keywords- Aerosol; ice nuclei; Weather modification; Reagent INTRODUCTION Ice forming nuclei activity is known strongly depends upon the temperature and the humidity of the environment. The choice of reagents for the weather modification purposes is mainly connected with temperature limitations. The very few reagents with extrimely high temperature thresholds were still discovered. But only silver iodide and copper acetylacetonate is known are able to provide up to 10 in the 14th power ice crystalls by one gramm of reagent at cloud temperature equal to minus 10 degrees centigrate. TABLE - Characteristics of ice forming reagents N
Compound
A. Inorganic comoounds Silver iodide Lead iodide Copper sulphide
Chemical formula
Phisicai & chemical properties DenSolubility (g) in sity g/cm 3 100g of H20 at+20~
Output of active particles per 1 g at-10~ Most Maximum probable obtained (natural (Lab.) conditions)
Notes
v
AgI PbI 2 CuS
5"67 6"16 4"6
3x10 "7 7x10 "2 practically insoluble
326
5x1015 2x1013 10 t2
1013 1012 l0 li
It is not to be dispersed thermally; used as a powder
Ice nuclei modification by variation of aerosol particles composition Compound
Chemical formula
Phisical & chemical properties Den- Solubility sity (g) in 100g g/cm 3 of H20
at+20~ B. Organi~ comoounds Phloroglucinol Metaldehyde
C6H3(OH)3 2H20 (CH3CHO)4
6
1,5-dioxinaphthalen
CIoH6(OH)2
7
Copper AcetylAcetonate
Cu(C~HTO~)2
1"45
14
1"13
Output of active particles per 1 g at-10~ Most Maximum probable obtained (natural (Lab.) conditions)
327
Notes
1013
insoluble
5x1013
Increased volatility restricts its application for remote seeding
difficult to disolve
4x1013
High efficiency is obtained only in a special generator with a supersonic nozzle. In a usual generator 10~2 g -1 at - 10~
up to 1014
Effective at supersaturation in a cloud
insoluble
One can't get this extrimely high yield by using the pure silver iodide (Gravenhorst et al., 1972). The impurities present in commercial reagent nevertheless make silver iodide the mostly used reagent for weather modification. The ice nucleating properties of copper acetylacetonate Cu(CsH702)2, with a high threshold temperature ..2~ (Malkina et al., 1978) was primarily investigated for the chemically pure substance. It was found that water supersaturation is necessary for its activation. The yield of IN with chemically pure reagent at water saturation is almost two orders less then that of at high saturation (Aksenov et al., 1980). The conclusion was made that the need of supersaturation for IN activation is the property of reagent. It has been experimentally determined that IN of copper acetylacetonate are steady to ultraviolet radiation. RESULTS Based on the recent experimental results we can draw the conclusion that the directed ice nuclei modification by variation of aerosol particles composition provides the activation of IN in the same extent as high supersaturation does. The main problem for copper acetylacetonate particles modification was the choice of chemically indifferent addents. We have found that urea satisfies to our purposes. It is hygroscopic and indifferent to reagent. The small amount of copper acetylacetonate and urea mixture was vaporized in the dry aerosol chamber. The
328
Malkina et al.
aerosol sample was then introduced into the cloud chamber, the appearing ice crystalls were counted and IN yield from one gramm of copper acetylacetonate calculated. Pure copper acetylacetonate yield was measured de the same way. In the presense of hygroscopic urea IN yield for copper acetylacetonate increases almost 50 times. This value was achieved when urea to reagent mass ratio equals 3. Higher ratios did not change the effect of sublimation. In this way, hygroscopic additives remove the dependence of activity of copper acetylacetonate on supersaturation of water vapor (Kim et al., 1986). Another possible additive was found: it was a successful combination of copper acetylacetonate with copper complex of acetoacetic ether. The yield of nuclei with the mix of copper acetylacetonate and copper complex of acetoacetic ether in saturated water vapor was 10 to 20 times as high as when pure copper acetylacetonate was used (Molotkova et al., 1991). The search for reactants called for developing a new method for express evaluation of nucleative activity of flashed substances, which could be of organic or non-organic nature. The known method based on counting the fallout of ice crystals using a microscope ensures sufficient precision, but takes a rather long time, and imposes certain requirements on the design of the refrigerator. We developed an express method for evaluating the nucleative activity: a light source was attached to the outside of the glass door of Grunland refrigerator. The light beam crossed the chamber diagonally and was bounced back by a mirror fixed to the back wall to a photocell outside the glass door. To prevent fogging, the mirror was gently warmed by electric heater. This double-beam techique in the chamber filled with supercooled water mist at a fixed temperature of-10~ allowed us to detect the fallout of the first, virtually invisible, ice crystals after instantaneous flashing of the sample in the middle of the chamber with the aid of specially designed electric device. The photocell, connected to a strip-chart recorder, detected changes of the fog density in the chamber. The yield of nuclei was compared against the yield of an equal-weight of silver iodide. The use of this express analysis allowed us to study a great variety of organic substances, acetylacetonates of metals and their mixture (figure). Earlier investigations (Patrikeev et al. 1975-76) showed that various additives, such as acetylacetonates of cobalt (Co), molybdenum (Mo), vanadium (V), also ensure high yield of nuclei. These additives have been used for seeding the clouds in those areas. Where the soil is low on such microelements. Well known is the disease of thyroid gland caused by deficit of iodine. Less widely known is the sickness caused by the deficite of cobalt, which is essential for hematogenic vitamin B~2 (it strikes goats, sheep, cattle, and is known a "bush disease", or "anaemia"). Thus, we discovered and studied new nucleative reactans (pure substances and mixtures) which, in addition to their straightforward purpose-discharging super cooled clouds, preventing hailstorms, dispelling smogs, fighting forest fires, help to improve the environment by making up for the shortage of microelements. REFERENCES Gravenhorst, G., Corrin, M.L. (1972), J.Rech.Atmos., Vol.6, N 1-3, p.205-212. Malkina, A.D., Patrikeev, V.V. (1978), Trudy CAO, Vo1.132, 103-107. (In Russian). Aksenov, M.Ya., Bromberg, A.V., Bitschkov, N.V., Kordyukevich, N.G., Plaude, N.O. (1980), Trudy CAO, Vo1.142, 82-88. (In Russian). Kim, N.C., Shkodkin, A.V. (1986), J.Meteorology and Gidrology, Vol.2, p.28-31. Molotkova, I.A., Pershina, T.A. (1991), Trudy GGO, Vol.534, 44-48. (In Russian). Patrikeev, V.V., Malkina, F.D., Kartsivadze, A.J. (1975), USA Patent, CI. 252-319, N 3887580. Patrikeev, V.V., Malkina A.D., Kartsivadze, A.J. (1976), FRG Patent N 2228281. We believe that the results of copper acetylacetonate ice nuclei modification by variation of aerosol particles composition will lead to its more effective use in weather modification programs.
Ice nuclei modification by variation of aerosol panicles composition
329
Figure The refrigerator c h a m b e r 1 - refrigerator, 2 - door, 3 - lighter, 4 - mirror, 5 - fotoelement, 6 - heater, 7 - sleam pipelene, 8 sample glass, 9 - refrigerator c h a m b e r equipment.
A
STUDY ON THE OF ICE NUCLEI IN BEIJING OF
L A I G U A N G X I A N G G U O
C L I M A T E C H A N G E C O N C E N T R A T I O N 1 963 TO 1 995
YOU W A N G
S H A O Z H O N G J I A X I O N G PI
YANG
C h i n e s e A c a d e m y of M e t e o r o l o g i c a l S c i e n i c e s AbstractThe concentration of ice nuclei activized in the temperature range from -15 to -30~ have been measured using a Bigg type mixing cloud chamber during the period from 18 March to 30 April 1995 at the western suburb in Beijing. Some variation characteristics are analyzed and the results are compared with the results obtained in 1963 w i t h the same observation condition. K e y w o r d s - Ice n u c l e i ; D u s t s t o r m ; P o l l u t i o n ;
Climate change
INTRODUCTION
A l a r g e n u m b e r of o b s e r v a t i o n s show t h a t t h e CCN c h a n g e s m i g h t e f f e c t e d t h e cloud r a d i a t i v e p r o p e r t i e s by w a y Of t h e p o t e n t i a l i n f l u e n c e of m a n - m a d e CCN on cloud a l b e d o a n d h e n c e in t h e c l i m a t e f e e d b a c k . T h e - I N h a s t h e s a m e i m p o r t a n c e as t h e CCN in m a n y a t m o s p h e r i c p r o c e s s e s (Vali, 1991), b u t so f a r o n l y a few d a t a of IN c o n c e n t r a t i o n c a n be u s e d to e s t i m a t e t h e p o s s i b l e c l i m a t e c h a n g e of IN a n d t h e a n t h r o p o g e n i c e f f e c t on it. The IN c o n c e n t r a t i o n in Beijing d u r i n g t h e p e r i o d from 18 March to 30 April h a d b e e n o b s e r v e d w i t h a m i x i n g cloud c h a m b e r w i t h 3 l i t r e s v o l u m e in 1963(You a n d Shi, 1964). T h e cloud c h a m b e r v o l u m e , c o n s t r u c t i o n a n d o b s e r v i n g m e t h o d w e r e all of t h e s a m e w i t h t h a t of I s o n o e t a1(1959). The r e s u l t s were d i s c h s s e d w i t h r e f e r e n c e to t h e c o n d i t i o n s of d u s t s t o r m a n d p o l l u t i o n . It s h o w n t h a t b o t h t h e p o l l u t i o n a n d d u s t s t o r m a r e i m p o r t a n t IN s o u r c e s in B e r i n g a r e a . The c o r r e l a t i o n f u n c t i o n b e t w e e n v i s i b i l i t y V~.~0~km) a n d IN c o n c e n t r a t i o n N (1-) a t -20~ t e m p e r a t u r e h a d b e e n f i t t e d , as N = 1 7 . 7 6 VIS-" w i t h a - 0 . 9 3 7 6 of c o r r e l a t i o n c o e f f i c i e n t . Since 1 9 6 0 ' s f o l l o w i n g t h e c i t y e x p a n d i n g , t h e a v e r a g e v i s i b i l i t y h a v e b e e n d e c r e a s i n g from 23 km in 1963 to u n d e r 10 km in 1989 in Beijing A r e a . It i m p l i e d t h a t t h e IN c o n c e n t r a t i o n m a y i n c r e a s e d by s e v e r a l t i m e s in 1990'. In o r d e r to t e s t t h e c l i m a t e c h a n g e , ~ t h e IN c o n c e n t r a t i o n h a v e b e e n m e a s u r e d in 1995, u s i n g t h e s a m e d e v i c e , in t h e s a m e s e a s o n a n d a t t h e same p l a c e w i t h 1963, a d d i t i o n a l l y a f i l t e r - s t a t i c diffusion chamber method was u s e d to m e a s u r e t h e IN c o u n t s i m u l t a n e o u s l y . In t h i s p a p e r t h e r e l a t i v e i m p o r t a n c e of d u s t s t o r m of E a s t e r n A s i a a n d t h e p o l l u t i o n of c i t y as t h e IN s o u r c e s a r e a n a l y z e d a n d t h e p o s s i b l e c l i m a t e c h a n g e of IN is a l s o d i s c u s s e d . M o r e o v e r some major r e s u l t s of t h e o b s e r v a t i o n a r e c o m p a r e d w i t h t h e d a t a s e t a t o t h e r p l a c e s in t h e world. S O M E C H A R A C T E R I S T I C S O F IN C O N C E N T R A T I O N IN BEIJJING IN 1995 The measurements of IN concentration are carried out at local time 10 am and 15 p m (BST) every day during the period from 18 March to 30 April in 1995. The 86 data set of IN concentration (in the temperature range from -12 to -30~ are obtained using a Bigg's mixing cloud chamber with two methods of introducing the water vapour and introducing the fog droplets generated by an ultrasonic nebulizer, in addition 86 data set of millipore filter (0.45 ~m) samples processed by the static diffusion chamber are also obtained.
330
Climate change of ice nuclei concentration in Beijing of 1963 and 1995
331
F i g u r e 1 i l l u s t r a t e s a d a y to d a y v a r i a t i o n in t h e IN c o n c e n t r a t i o n o b s e r v e d w i t h t h r e e m e t h o d s in t h e p e r i o d from 18 March to 30 April in 1995. It s h o w s t h a t t h e p e a k s or v a l l e i e s of IN c o n c e n t r a t i o n v a l u e s on t h r e e c u r v e s a p p e a r a t s a m e time b a s i c a l l y . The IN c o n c e n t r a t i o n o b s e r v e d by i n t r o d u c i n g fog d r o p l e t a r e h i g h e r 1 - 2 t i m e s t h a n t h e i n t r o d u c i n g v a p o u r in m i x i n g cloud c h a m b e r . T h e r a t i o of ice n u c l e i c o n c e n t r a t i o n o b s e r v e d by m i x i n g cloud c h a m b e r to t h e d i f f u s i o n c h a m b e r a r e in t h e r a n g e from s e v e r a l to o v e r 1000, a n d i n c r e a s e s w i t h t h e m e a s u r i n g v a l u e of Bigg's m i x i n g cloud c h a m b e r , t h e more t h e ice n u c l e i , t h e h i g h e r t h e r a t i o . F i g u r e 1 s h o w s t h a t t h e high IN c o n c e n t r a t i o n v a l u e s r a i s e d a t t h e d a y s w i t h h e a v y p o l l u t i o n on 20 March and d u s t s t o r m on 6 April. T h i s r e s u l t s h o w s t h a t b o t h of t h e c i t y p o l l u t i o n a n d t h e d u s t s t o r m a r e s t i l l t h e i m p o r t a n t IN s o u r c e s as in 1963. RI ( L -I ) 104
103 ltle 102
10
1
10-1
10_2
tJlJl
I Ill
18 20 1995,3
illllll
25
t I111
}0 199~, 4
5
i1111
10
1 tl
Ill
15
1ill
20
llJ|llll
25
30
Figure i The variation of IN concentration at -20~ day by day, NI~ --observed by mixing cloud chamber introducing fog droplet, NIv--by introducing water vapour, Nil-observed by filter-diffusion chamber with ice supersaturation Si=0.21. F i g u r e 2 s h o w s t h e IN c o n c e n t r a t i o n o b s e r v e d in Beijing, N a g o y a a n d Spain. It can be s e e n t h a t t h e IN c o n c e n t r a t i o n s in Beijing a r e h i g h e r t h a n a t o t h e r p l a c e s , e s p e c i a l l y in t h e w a r m e r t e m p e r a t u r e . The IN c o u n t o b s e r v e d by f i l t e r - d i f f u s i o n chamb.~r in Beijing a t -20~ w i t h w a t e r s a t u r a t i o n c o n d i t i o n a r e in t h e s a m e r a n g e (0.1 to 10 1-) with t h a t o b s e r v e d by R o s i n s k i et a1(1995) u s i n g s a m e m e t h o d o v e r t h e E a s t C h i n a Sea. THE CLIMATE CHANGE OF IN CONCENTRATION
The data set of IN concentration obtained in have 1995 compared with that in 1963. It shows that the mean IN concentration in 1995 at -15, -20, -25 and -30~ have increased by 20, 15, i i and 6 times, respectively. Table l Summarizes the values of
You et al.
332
parameter A and b in different years and different weather situations. It indicates that the IN counts have greatly increased for past 32 years and the anthropogenic of IN source have more importance than the natural source in the western suburb of Beijing, especially for the ice nuclei active at the lower supercooled temperature (T>-24~ An analysis based on the results of Rosinski et al ( 1 9 9 5 ) s h o w s that the IN concentration over East China Sea under the air masses reached the sampling station from South China are higher than that from desert of Northwest China. It may be as a collateral evidence to
illustrate the importance a n t h r o p o g e n i c sources. THE RELATIONSHIP CONCENTRATION SITUATION
BETWEEN WITH
~, (i-I ) 10 4
i
1~
I
1
~
of
T H E IN WEATHER
Figure 3 shows the variations of IN concentration and the surface pressure day by day. It can be seen that all of the high IN count appeared on the days with low pressure. Usually the peak values of IN concentration were corresponding to the lowest pressure. This might be owing to that these situations are advantageous to concentrate pollutant or to lift the dust particle in the convergence zode of depression. The ratio of IN concentration observed by mixing chamber to that by diffusion chamber in the heavy pollution air and in the dust storm are estimated, the ratio values are 1000 and 60 for
I:
_ 0.2
-
. : !
r
-30
Isono perez
, I
et et
,
I
-25
(1959)
al al
, i
(1985) I
I
I
-2o
i
I
I
, I
-15
\
,~i
('c)
Figure 2 The a c t i v i t y s p e c t r a of IN o b s e r v e d using mixing cloud chamber at d i f f e r e n t time and by d i f f e r e n t a u t h o r s , c u r v e 1995a o b s e r v e d in Beijing - introducing fog d r o p l e t s into cloud chamber, 1995b--introducing water v a p o u r o b s e r v e d in Beijing, 1 9 6 3 b - - s a m e as 1995b but in 1963, I - - o b s e r v e d by Isono et a1(1959) in J a p a n u n d e r the air masses r e a c h e d from North China, P - o b s e r v e d by Perez et al in Spain(1985).
Table 1 The p a r a m e t e r A and b of N=A Exp(bT) Note
A(1 -I)
b(~ -1)
A v e r a g e for 1995
0.1764
0.~143
A v e r a g e for 1963
0.00106
0.42
The dust storm on 6 April 1995
7.180
0.2085
The h e a v y pollution air on 20 March 1995
35.79
0.1426
Climate change of ice nuclei concentration in Beijing of 1963 and 1995 ( L -~ )
c, D
102
I
10
,L
10301:?(hPa~
J
1
I
P'
20
1020 [
/2~ la b
AI
'
I
' '
I
I
II
,
30 d l ~
I
I
5
C
, 15
102
, z,.;.
(-2o"
,
20 /~
[, f I I
F
,
I
e
'/i/ I
j
1
10
;
E
I I
, 25
,
10
J,
30
"-! 1030 ~ P(hPa)
h | I
-:1 1020 (hP
1010 1000 P(hPa )
333
1010 1000
20
3/95
25
Fig.3 The v a r i a t i o n s p r e s s u r e in Beijing.
30
4/95
5
10
d a y by d a y
~5
2o
of IN c o n c e n t r a t i o n
25
(-20~
30
and
surface
heavy pollution and dust storm respectively. Probably it reflects that the aerosol particles have larger size and less solubility in dust storm than that in pollution air, so the activity of aerosol particles as sublimation nuclei in dust storm are higher than that in pollution air, and the activity as freezing nucleation are lower than that in pollution air. CONCLUSION The results of IN count observation during the spring in 1995 at Beijing are summarized as follows: I) The IN concentration in the temperature range from -15 to -30~ have increased by 20 to 6 times in the past 32 years. 2) The IN produced by the anthropogenic source have more important than the natural source in the suburb of Beijing. 3) Usually the higher IN concentration appeared in depression belt. 4) The ratio of IN count observed by mixing cloud chamber to diffusion chamber are from several to i000.
A c k n o e l e d g e : This s t u d y is s u p p o r t e d
by N a t u r a l S c i e n c e s F o u n d a t i o n of China. REFERENCES
Isono, K., et al, (1959), J. Met. Soc. J a p a n , Set II, 87(6), 2 1 1 - 2 3 3 . ~'erez, P.J. et al. (1985), J. Rech. Atmos., 1 9 , 1 5 3 - 1 5 8 . Rosinski, J., et al, (1995), Atmos. R e s e a t . 36, 9 5 - 1 0 5 . Vali, G., (1991), WMO/TD-No.423. You, L. G. & A n y i n g , Shi, (1964), Acta. Met. Sinica, 34, 5 4 8 - 5 5 4 .
ON
VSit'2ICAL
Cs
OF
,S(.~Nd ' :"' :. :i "O z.L~.z " i", . ; _ ~ : j
~ICL;I'~S AND A~ROSOL P ~ ' R':' ifi
0~.,.
C ' N~iCLEI
.ai%ioSP'-l";~ '
V.G.Ki~orgu,:k~-~i, Z.V.ii,,vedelidse Geophy~;ics institute,
xcadel,ly 02 ~ciences
:ibilisJ,
Abstract
-2he
the v e r t i c a l
rticles over
the
(AP)
oi" the aircraft
of the ice nuclei
substrate
in the p r e m o u n t a i n
region
height
and
more
creased
vertical
nuc!ai
the
in a t m o s p h e r e
aerosol
the
has very
substrate
effect
carried
rt of main C a u c a s u s
Range
at the
over the B l a c k and in the
stable
taken up to the
a!lowimS
filter
chamber
error
is
but
at
.... m e a s u r e m e n t s 20~
it
equals
to
the
Then
20~6.
334
effect.
these
The m e a s u -
during
the
of atmosphere.
by the
"Grad-3" largest
L!ultiple m e a s u -
in the premo5uutain pa-
region
the p a r t i c l e
in the t h e r m a l - d i f f u s i o n of
stepp
this
stratification
as well.
J.,~e~ulwhiil_e, in the
condk~cted in a t m o s p h e r e
sea,
5 km height
to f r a c t i o n a t e
taken bj m e m b r a n e
were
in order to i d e n t i f y
out
and ice
~.~ature uncle-~ the effect
is irievitable.
were
pactor
is in-
partio!es
and clokd~iess.
rements
were
and
with
con~centration
aerosol
complicated
of the IN and AP C o n c e n t r a t i o n
samples
that
stepp
fraction
partJ_c!es,
rements
weather
to the
pa-
fraction
is d e c r e a s e d
nuclei
of the n a t u r a l
stratii'ication
by a i r c r a f t - l a b o r a t o r y
cyclonic
nuclei
It is shown
in c o m p a r i s o n
the active
Ice nuclei,
distriOution
layer
the active
are given.
on
height.
of the a t m o s p h e r i c
bou~dary
investiga;ions
(ii'r) ~id aerosol
IN and ~2 c o n c e n t r a t i o n
slowly
with
keywords-
Georgia
ooncentrat:Lon and
different
sea r e g i o n
The
results
change
of Georgia,
three
sine.
The
samples
cascade
~nd
im-
sal,:iples were
were
developed
(i'@norsu.ani, 1984).
-at -6~
anti-
The
equals
to
The 10096
On vertical change of concentration of aerosol particles and ice nuclei ~ig.1 seen
times
gives
l~J a~Id AP :~ore
profiles
comcentratioi~
la~'ger ~i~a:l over
h eigl~t
rear
the v e r t i c a l
quicker.
the effect
the
The
of the
in the permoLur~ain
sea and
range
stepp
H, KM
of IN arld hs stepp
effect
and
and
I
i
I
2
contrasting, the
shell
same
have
at the
change
particles
(b) over
by the w a t e r
ble particles.
shell
of p a r t i c l e
in the
increase
the v e r t i c a l
and d=O.2~m.
It is seen
vity
of n a t u r a l
aerosols
clei
is i n c r e a s e d
change
with h e i g h t
to 5 km is increased,
from
(a) and
of aerosol
differe~r~
:regions.
region,
over
considerable
the
of the
are
the
at
coated
solu-
ice-forl~iins actisizes
concentration for both
by --an order
whereas
sea bec~ne
amount
for m e a n - c u b i c cases
showed
region
and do mot have
humidity,
of the r e l a t i v e
the r e l a t i v e
3-stepp.
taken by i m p a c t o r
nonsoluble
in any
concentration
the p r e m o u n t a i n
taken
approximately
I
I
of the r e l a t i v e
at -20~
that
the
s~nples
angles,
contained
Fig. 2 shows
the
v/he-
to,o 31L t.
of v e r t i c a l
at -20~
the p a r t i c l e s
and
I
2-premountai~
samples
the acute
conditions
with
Ti~e e l e c t r o n i c -
I
,,,
1-Blac!: sea,
water
II
6,0
FiS. I The
analysis
is in several
~ Km i ~ e i g h t
less.
it
I\\
I
of ice nuclei
the p a r t i c l e s
to
As
ox.Q o~
2,0
that
region
up
sea is much
\
microscopic
concentration.
is d e c r e a s e d
occurs
335
of d=10~m
os ice nu-
sizes
and up
of m a g n i t u d e .
336
Khorguani and Khvedelidze
Sesides, the 1"faction of the active particles for d:iOt~m is approximately
induced,
by an order of magnitude at other equal
on the ice-s
of the increase
different.
ration,
taking
reactions ation the
of the active
related
First
account
particles
takes place
~,WM
size influence
(}ihorguani,1990).
particle
It was
The reasons
traction with height may be
of all it car~ be caused by the photo-chemical
to the increase
into
than for d-O.2~m.
by the particle
os aerosol
vfith height
tha~
of the osone
o~:one ? r o r r ; o ~ e s
(O;~'irishvi!i, Kharchii'ava,1977).
soluble
marion
conditions,
activity
quite
greater
(conde~:so~tion
the
concei~t-
ice-!:~hase
form-
Apgarently , the wash-out
nuclei)
and the ice-formation
on w h i c h
does not
the
cloud
occur.
os for-
prob-
This
S"
///
3-
o
M
o
9
//
9
o--~ •
//
---3
o,~,e
e o x
n
L
f O - -~'
10 - 2
Fig.2 The vertical
ming activity
me'an-cubic l-Black
lem needs
profiles
of natural
diameter
sea,
!
j~ ,"
of the relative (a)-1Of~m;
3- premountain
.
I 0 - ~ ,N.....i
aero'~'ols at -20~
of particles:
2-stepp,
II
t O " 5"
ice-for-
and for
region.
(b)-O.2]~m.
further investigations. REFERENCES
G:~irl.si~vili '2.G. and itharchilava J.F. oso~~e i n
~be possible
ti~e ~ h e r ~ o m e n o n o2 i c e - f o r m a t i o n
Atmospheric
iihorguarli V.@.
(1977)0n
wth. l{idrometeoisdat,
of
i~l ti~e c l o u d s .
and Oceanic Physics. 13, I, 100-102.
(19U4) ivLicrophysics of hailstone
role
generation
(in i~ussian) and gro-
i.ioscow, IS6 pp.(in Russian)
On vertical change of concentration of aerosol particles and ice nuclei ~{horguani V.G.
(1990) Ice-forming
Aero~ois, Japan,
7T
,10
. . . .
..J
.-p_ E t--
90-
(RH/%-f)*ln 1( i Life
",~
time/s
=
exp
--
g
80-
f g R
tD st"
70- o
60
0
SO0
~-
1000
54.110844185 16.067506225 0.90752594806
i
1 SO0
i
2000
t
2500
I
3000
3500
Life time of the contrail/s
Figure 3" Scatter plot of the pairs of ambient humidity and simulated contrail life time for the threshold values for faint contrails. Exponential function fitted to these data. CONCLUSIONS In this paper a cloud scheme is exercised in a meso-scale environment with moist processes generating a contrail in a region with subsaturated air. The simulated ice concentrations are derived from the mixing ratios and number concentrations of the water species. The parameterizations lead to a time of persistence of the contrail of about 10 minutes. The general feature of the contrail (ice m ~ s and concentration) seems to be well reproduced with the cloud scheme. The simulation results indicate an exponential variation of the life time of the contrail with changes in the moisture of the ambient air. Future studies will examine the possibility to extrapolate this relation to different atmospheric conditions.
References I(apitza, H., and D. P. Eppel, The non-hydrostatic mesoscale model GESIMA. Part I: Dynamical equations and tests, Contr. Atmos. Phys., 65,129-145, 1992. Levkov, L., B. Rockel, H. I(apitza, and E. Raschke, 3D meso-scale numerical studies of cirrus and stratus clouds by their time and space evolution, Contr. Atmos. Phys., 65, 35-58, 1992. Hennings, D., M. Quante, and R. Sefzig, International Cirrus Ezperirncnt i989, Field Phase Report, University of Cologne, 1990. Albers, F., M. Quante, and E. R~chke, Aircraft measurements in high altitute contrails during ICE 1989, 7th Conference on Atmospheric Radiation, 23th-27th July 1990, San Francisco, 1990.
MONTE
CARLO
STUDIES OF WATER/ICE ADSORBED EFFECTS OF LATTICE SHIFT
ON MODEL
AGI:
B. N. HALE and D. J. DIMATTIO
Department of Physics and Cloud and Aerosol Science Laboratory, University of Missouri-Rolla, Rolla, MO 65401 USA
Abstract - Monte Carlo simulations employing the Bennett technique for calculating free energy differences are used to study the effect on small adsorbed H~O clusters of lattice shift in an underlying model hexagonal AgI substrate. A periodic array of Ag + and I ions is used to model the substrate and the revised central force potentials (RSL2) are used for the HzO-H20 interactions. Free energy differences are calculated for the adsorbed water systems with differing model AgI basal face lattice constant. Effects of temperature and cluster size are also considered. Keywords - Water; AgI; H 2 O; Nucleation; Monte Carlo; Free energy, Adsorption; Molecular; Effective potentials; Substrate; Lattice mismatch. INTRODUCTION Recently, Zapadinsky and Kulmala (1995) examined the shift in Helmholtz free energy, AF(n), for adsorbed n molecule water clusters on coherent substrates with stretched lattice constants at T - 130 K. Monte Carlo simulations of small model clusters were performed, using a Bennett (1976) technique for determining Helmholtz free energy differences. The purpose of the study was to examine the dependence of the energy shifts on cluster size, n, and lattice mismatch, (a-ao), where ao corresponds the basal face lattice constant of pure ice I h . Cabrera (1965) had proposed that AG/n=AF/n =(a-ao)2/,/n, whereas Tumbull and Vonnegut (1952) had used AG/n - (a-ao)2. The work of Zapadinsky and Kulmala (1995) found that AF/n increased with n and was consistent with a (a-ao)2 dependence. As noted by Zapadinsky (1995) the potentials used were formulated to account for the distortion of hydrogen bonds in crystal hydrates (Efimov et al. 1984) and had not been tested to describe other macroscopic properties of ice. In particular, the potentials were intended to treat ice substrate systems which were only slightly perturbed from the low temperature ice structure. In previous work we developed a model of the same system with effective water-substrate atomatom potentials (Hale et al. 1981, Taylor and Hale, 1993) and RSL2 water-water potentials (Rahman and Stillinger, 1978) applicable to arbitrary H20 molecular structure on the model AgI basal face substrate. RSL2 potentials also support a stable bulk ice Ih system at low temperatures (with basal face lattice constant, a~c~ = 4.54/~ and O-O separation distance, R o= 2.78 ,~, near 100 K) which melts near 280 K (Han and Hale, 1992). At higher temperatures (near 280 K) the R o is closer to 2.8 A (Han 1989). The substrate Monte Carlo studies (Taylor and Hale 1993) simulated a periodic water/ice monolayer on the model AgI substrate at nine temperatures and examined the H20 layer structure and state as a function of T. It was found that the first layer of H20 on the substrate was solid -like and highly structured in five and six membered tings centered on exposed iodine atoms up to temperatures of 325 K. This model water-substrate system appears to offer an opportunity to extend the studies of Zapadinsky and Kulmala (1995) on the effect of lattice shift. Below we report the Helmholtz free energy 349
350
Hale and Dimattio
shifts for two cluster sizes (n = 6 and 24) and two temperatures, (T = 130 K and 280 K) as function of (a-a3 in our model system. The inclusion of data at 280 K is included to provide some information about the systems at temperatures near the model ice melting point where processes applicable to atmospheric ice formation occur. THE POTENTIAL MODEL The system consists of n water molecules on the model basal AgI substrate. The internally rigid water molecules interact via the RSL2 and have intramolecular HOH angle and OH bond length equal to 104.5 ~ and 0.972 ,&, respectively. The RSL2 potentials assume a three point charge model for water with H and O effective charges equal to q = 0.32983e and -2q, respectively. In the present simulations a cutoff of 6 ,& is used for the water-water potentials based on oxygen-oxygen separation distance. The substrate model consists of a rigid periodic army of Ag and I atoms in a wurtzite structure with an iodine-exposed basal face, and a basal face lattice constant, a. The "ideal" lattice constant is chosen to be ao = 4.58/~. The latter constant corresponds to the experimental value for AgI at temperatures near 273 K and is close to the a~c, near 280 K. For purposes of comparison, this value of ao is used for both temperatures. The Ag and I atoms interact with the atoms in the water molecule via effective Coulomb potentials, Lennard-Jones cores and polarization terms. Effective charges for the Ag and I are ~:0.4e (Hale et al., 1980) and a hard wall 1,~,above the plane of exposed iodide atoms (z=0) is assumed to prevent H20 molecules from channeling into the substrate. When the substrate lattice is stretched (or compressed) the effective surface charge density on the top (iodine exposed) layer of substrate atoms is altered; this can mask the effect of the lattice shift on the free energy, particularly for large shifts. In order to preserve the overall strength of the substrate-water interaction, the effective charge on the Ag and I ions is scaled with the shift in lattice constant (a-ao) to keep the surface charge density constant.
COMPUTING TECHNIQUES Metropolis Monte Carlo simulations (Metropolis et al., 1953) are executed at two temperatures, T = 130 K and 280 K, for two water cluster sizes, n = 6 and 24 molecules. In this Bennett method each cluster is simulated in two systems: on the ideal substrate with ao = 4.58 A (called system 0) and on the substrate with a*ao (called system 1). But for each system the total potential energy can be calculated for both substrates (and their difference, AU averaged) even though only the system substrate affects the positions of the water molecules. The AU -- U(a) - U(ao ), is binned to form normalized probability distributions, PI=,,I(AU) for the two systems. Using the method of Bennett (1978), one obtains the Helmholtz free energy difference, AF = F(a) - F(ao), between the two systems from PI(AU)/Po(AU) = exp[(AF - AU)/kT]
(1)
In practice one plots histograms of P1 and P0 vs AU and uses the overlap region of the two histograms to generate an average AF ( Zapadinsky and Kulmala, 1995). Each cluster is equilibrated for n million steps and followed by the binning of AU for n million additional steps. Also calculated are differences of the average system potential energies, < U(a) > i < U(ao) > o and structure factors for RSL2 ice Ih, S(~=1,2) with reciprocal lattice vectors ~ = (2/3) '~ Ro[~ + (-1)i3(-'h)S'] and Ro = 2.78 ,~, (Han 1992). The S(k3 are given by: S(k.,) = (1/n 2 ) ( I Ej=I, n exp(ik(rj ) I 2 )
(2)
Monte Carlo studies of water~ice adsorbed on model AgI
351
RESULTS AND CONCLUSIONS In Fig. 1 are plots of AF/(nkT), [ < U ( a ) > ~ - o]/(nkT), S,vo-1/2[ S(kl)+ S(k2)] for two temperatures, T = 130K and 280 K, vs lattice constant shift, (a-ao). On the left (in Figs. la, lc, le) are the results for the n = 24 molecule water cluster and on the right (in Figs. l b, ld, and 1f) are results for the n = 6 cluster. In all cases we find that AF has a minimum near ao = 4.58 A (R o = 2.81 ,~,) and that the dependence of AF on lattice shift is consistent with AF ~ (a-ao). The latter dependence differs from that of Zapadinsky and Kulmala (1995) whose results are closer to a (a-ao)2 dependence. We note, however, that they did not have available a mechanism to scale the surface charge density with lattice shift. With respect to the n dependence, we find at the low temperature (130 K) AF increases with n (in agreement with Zapadinsky and Kulmala). At the high temperature (280 K) we find that AF decreases with n. This suggests that at higher temperatures entropy plays a larger role in minimizing F. However, our sample of cluster sizes is limited, and perhaps biased by use of n = 6. As is evident in the differences in the average potential energies for the system (Figs. lc and ld) for no substrate lattice constant studied did we find a substrate which provided a lower average system potential energy than the "ideal" substrate with a = ao = 4.58 A. The magnitude of the A < U/(nkT) > -[ < U(a) > 1 - < U(ao) > o ]/(nkT) is also about two orders of magnitude larger than AF so that the entropy differences, AS/(nk), are comparable to A < U/(nkT) > . This underscores the nonnegligible role which entropy plays in determining free energies of formation for such clusters on ice nucleating substrates. The structure factors for the systems (Figs. le and If) indicate that the ice Ih basal face structure is maximal in the region where AF is minimal. This is more pronounced at the lower temperature (130 K) and for the larger cluster (n = 24). At the higher temperature (280 K) the cluster structure appears less sensitive to the underlying lattice constant. Work is in progress on a similar study of the effect of substrate lattice constant shifts on adsorbed water layers with periodic unit cells. ACKNOWLEDGMENTS This work was supported in part by National Science Foundation under Grant No. ATM93-07318. We would like to thank E. Zapadinsky and M. Kulmala for helpful discussions about this study. REFERENCES Bennett, C.H. (1976)J. Comput. Phys. 22, 245. Efimov,Y and Gorbunov,B. (1984) Kristallograph. 29, 849. Hale, B.N., Kiefer, J. (1980) J. Chem. Phys. 73,923. Hale, B.N., Kiefer, J., Terrazas, S. and Ward, R.C. (1980)J. Chem. Phys. 84, 1473. Hale, 13.N., Kiefer, J. and Ward, C.A. (1981) J. Chem. Phys. 75, 1991. Han, K. (1989) "Monte Carlo Study of Melting of a Model Bulk Ice" Ph.D. Thesis, University of Missouri-Rolla. Han,K. and Hale, B.N. (1992) Phys. Rev. B. 45, 29. Stillinger, F.H. and Rahman, A. (1978)J. Chem. Phys. 68,666. Taylor, J. and Hale, B.N. (1993) Phys. Rev. B. 47, 9732. Ward, R.C., Hale, B.N. and Terrazas, S. (1982)J. Chem. Phys. 78,420. Ward, R.C., Holdman, J.M. and Hale, B.N. (1982) J. Chem. Phys. 77, 3198. Zapadinsky, E.L. and Kulmala, M. (1995)J. Chem. Phys. 102, 6858. Zapadinsky, E.L., Gorbunov, B., Voloshin, V. and Kulmala, M. (1994) J. Colloid Interface Sci. 166, 286.
Hale and Dimattio
352
b--,
0.07
-
0.05
-
0.03
-
. 1. Both of them correspond to the equilibrium of an embryo with water vapour. The radius of the equilibrium embryo can be found from the equation (dAG/dre =0):
S~=awexp
where m =
(Crg,-
~
fX22
= :r R
IOeg M ~ dre IdSeg dSesll RgT p-~ dVe\ dr -m dr ) '
(4)
and
O'es)/O-eg
g5
+2(1-mX)g X2 + (1-m2)[g3 1 -
(m-X)(1-mX)Xl}g2 9
(s)
These formulas allow to obtain values of Nw and re, which correspond to equilibrium embryos. Equilibrium supersaturation was calculated using the expression (4) in the case of mixed particles that consist of spherical insoluble core and NaCI (10 ~6 g). The activity coefficient of water was taken from Pmppacher and Klett (1980). It was assumed that the volume of solution is equal to the sum of volumes of water and salt. Dependence of equilibrium supersaturation on the embryo radius is similar to dependence that was obtained in the case of soluble substances only, see Figure 1. Insoluble substances affect equilibrium supersaturation considerably. Equilibrium supersaturation is influenced by the contact angle. An increase in the cosine of the contact angle from -1 to 0.7 leads to reduction in supersaturation (in terms of Sw- 1) by more than 3 times. It was shown that in the case of greater particles that contain 1015 g of sodium chlorine the effect is stronger. An increase in the cosine of the contact angle from -1 to 0.9 leads to reduction in supersaturation by more than 5 times. Equilibrium supersaturation is influenced by the radius of insoluble core. Change of the radius of the insoluble core from 10g to 3 10.4 cm leads to reduction in equilibrium supersaturation by 3 times (Figure 2). Thus, both size and cosine of the contact angle of the insoluble core affect equilibrium supersaturation considerably. The Kohler theory corresponds the case of the cosine of the contact angle m = -1 (curve 1 in Figure 1). The difference in values of equilibrium supersaturation (S,,- 1) calculated with Kohler theory and suggested approach can be as much as 5 times. Thus, Kohler theory cannot be recommended for calculation of equilibrium supersaturation in the case of atmospheric aerosols that contain both soluble and insoluble substances.
Water nucleation on aerosol particles containing soluble and insoluble substances
409
LIST OF SYMBOLS Water activity in the solution Change in the Gibbs free energy as a result of an embryo formation from Nw moles of water Contact angle of an embryo Molecular weights of water A number of moles of water Radius of an embryo that contains ]v~ moles of water Radius of an embryo Radius of the insoluble core of an aerosol panicle The universal gas constant Density of water Supersaturation of water vapour above a flat surface of pure water Area of the interface between an embryo and substrate Area o f the interface between an embryo and gas phase The interfacial free energy of the border between the panicle surface and gas phase The interfacial free energy of the border between an embryo and gas phase The interfacial free energy of the border between an embryo and substrate Temperature
aw
AG(Nw) m M~ nw re r
R
R~ p~
& Ses Seg
Crg, O'eg O'es
T
(Sw-I) * I 0 0
0.2
-0.2
-0.4 1 0.2
1 0.4
1
1 0.6
1
1 0.8
1
1 1.0
_ radius
Figure 1. The dependence of the equilibrium supersaturation of water vapour on the radius of an embvo corresponding the extremums of tile free energy. T = 293,15. m, = 10~6 g (NaCl). The contact angle m = -1 for the curve 1, 0.3 for the curve 2 and 0.7 for the curve 3. R = 8 x 10 -5 cm.
Gorbunov and Hamilton
410 (Sw-I) *i00
0.4
0.2
-0.2
-0.4 1 .... 1
0.2
0.4
0.6
1
0.8
I
I
i. 0
radius
Figure 2. The dependence of the equilibrium supersaturation of water vapour on the radius of an embryo corresponding the extremums of the free energy. T = 293,15. a - m, - 10 -~6 g (NaCI). m = 0.7. R = 10 g for the curve 1, 3 x 10 .5 for the curve 2 and 3 x 10-4 cm for the curve 3.
REFERENCES Gorbunov B. and Hamilton R. (1996) Multicomponent nucleation on aerosol particles, which contain both soluble and insoluble substances. Proceedings of 14th international conference on nucleation and atmospheric aerosols, Helsinki, Finland; Elsevier. Kohler H. (1936) The nucleus in and the growth of hygroscopic droplets. Trans. Faraday Soc., v. 32, p. 1152. Pruppacher H.R. and Klett J.D. (1980) Microphysics of clouds and precipitation. Reider P.C., Holland, 714p.
TRANSPORT
OF RADIOACTIVE
MATERIALS
IN CONVECTIVE
CLOUDS
I. V A L K A M A a and R. POLL,g, N E N b
a Finnish Meteorological Institute, P.O.Box 503, FIN-00101 Helsinki, Finland b Finnish Centre for Radiation and Nuclear Safety, P.O.Box 14, FIN-00881 Helsinki, Finland
A b s t r a c t - After the Chernobyl accident large and highly radioactive particles were found
hundreds of kilometres from the plant. To explain these findings either the effective release height must have been higher than previously reported or convective flows lifted the materials upwards during transport. Present trajectory models do not sufficiently take into account the phenomena related to deep convection. A model for convective updraft is implemented into a trajectory model. Test calculations show that travel distances of the particles may change considerably if deep convection is included. In addition, the fallout area may be shifted and broadened.
INTRODUCTION Particles and gases released in a nuclear accident may be transported hundreds ofkilometres from the plant. Highly radioactive particles are of special importance. Their radiological risks are different from those of gaseous species. Even an individual nuclear fuel particle may represent a potential health hazard. Severe local injuries may be produced in a short time if nuclear fuel particles larger than a few tens of micrometres in aerodynamic diameter are deposited on the skin (P611~inen et al., 1996) Long-range transport is determined mainly by atmospheric conditions, properties of released material and effective release height. Particles and gaseous species are not transported similarly even if they are released simultaneously at the same effective release height. Owing to sedimentation, the large particles settle to lower air layers where atmospheric conditions may be different. Similarly, when particles encounter convective currents of warm air they may be lifted into altitudes with different wind conditions. Comparison between environmental findings of radioactive particles with trajectory calculations allows estimation of the effective release height if atmospheric conditions and particle properties are known (P611~anen et al., 1996). If, however, the particles are lifted up in convective currents the comparison is not straightforward. In the present study a model for convective updraft is implemented in the long-range trajectory, dispersion and dose model known as TRADOS (Valkama et al., 1995). Analyses previously performed suggest that in the Chernobyl accident convective cells with rising currents of warm air may have transported radioactive materials to altitudes higher than those reported earlier (Valkama et al., 1995; P611~inen et al., 1996). Here, trajectories with and without convective uplift are compared.
DEEP CONVECTION Atmospheric phenomena that mainly affect the long-range transport of gases and particles include wind speed, wind direction, depth of the atmospheric boundary layer and scavenging by precipitation. The top of the boundary layer, the entrainment zone, is characterized by the presence of a marked wind shear. When the growing mixed layer is topped by clouds the airborne species may become entrained into an environment of high water content, leading to efficient in-cloud scavenging and, perhaps, chemical reactions. The boundary layer itself may be a source of cloud formation.
411
412
Valkama a n d P611iinen
A zone of horizontal convergence is required to maintain a deep convective cloud; this zone can encompass distances 1--2 km outside the visible cloud. This inflow may draw airborne materials into the cloud; if this occurs near the source these materials can be lifted upwards, giving rise to an increase in effective release height. The rising currents inside a cumulonimbus cloud may attain velocities of tens of metres per second. In a cumulonimbus cloud, radioactive substances may be injected into the free atmosphere. TRADOS utilizes numeric weather data from the Nordic High Resolution Limited Area Model (HIRLAM). Special data, compiled by the Danish Meteorological Institute, are used in the present study covering the period of the Chernobyl accident (Valkama et al., 1995). The temporal resolution of the data is six hours and spatial resolution 55 km; deep convection, thus, is a subgrid scale phenomenon. Earlier attempts at modelling the updraft are based on subjective analyses of synoptic observations (Bonelli et al., 1991). This technique, requiring manual correction of particle height during transport, can be applied only to isolated cases. Full parametrization based on vertical potential temperature gradients can be formulated using the HIRLAM model level data, but this would require major changes in TRADOS; an empirical and simplified approach is used, instead. The evolution of convective conditions correlates with available thermal energy. The convectivity (lability) of the atmospheric boundary layer is estimated using a prognostic variable, h, as defined by George (1960) K : (T850hPa + Td,5ohe,, ) -- (T7oohPa+ TdToohpo) -- Tsoohe,'
where T is temperature and Ta dew-point temperature at constant pressure levels of 850, 700 and 500 hPa. This index has been extensively used in routine storm forecasting because it correlates well with the probability of thunderstorm occurrence. The value of K should be larger than 25 to ensure that only deep convection is considered in fire calculations. Since frontal convection is already included in the HIRLAM data, deep convection is allowed to occur only in daytime (solar elevation greater than 10 degrees). The vertical displacement inside one time step was set to 1000 m (Bonelli et al., 1991) and only one displacement is accepted in each trajectory. The procedure described above predicts the occurrence of convective cells in the Ukraine-Belarus region for almost every afternoon during April 26--May 10, 1986, which is supported by synoptic weather observations.
CONVECTIVE UPDRAFT IN THE CHERNOBYL ACCIDENT Travel distances of particles released in the Chernobyl accident are compared with calculated distances for the largest (most active) particles found at each site (POll~inen et al., 1996). Findings of large particles in Finland, Sweden, Lithuania and Poland suggest that the materials could have been released from considerably higher altitudes than previously reported (P611~en et al., 1996). Particles detected in Finland must have originated at least 1000 m above the ground (Valkama et al., 1995), otherwise they would not have reached Finland. Trajectories originated between 500 and 1100 m have entered Sweden. Possible deep convection may change these limits, as illustrated in Fig. 1 and 2. Air parcels from an effective release height of 700 m without experiencing convective uplift reach Sweden whereas after an updraft of 1000 m they reach Finland. During the early stages of the accident the possible deposition area of large particles was narrow up to 1000 km from the plant (P611~en et al., 1996). In addition to turbulent dispersion, convective clouds may broaden the deposition area by changing the transport direction. Particles in the lowest air layer (effective release height less than 400 m) are originally transported towards central Europe while convective uplift may transfer them to upper air layers flowing towards Scandinavia.
413
Transport of radioactive materials in convective clouds
Similar results are obtained by Bonelli et al. (1991) for the later stages of the accident. They showed that without deep convective phenomena the transport of radioactive materials from Chernobyl to Italy cannot be explained. Our preliminary calculations suggest that deep convection might also explain the findings of particles larger than 20 gm in aerodynamic diameter detected in Hungary, Romania, Bulgaria and Greece.
SWEDEN
t../ FINLAND
FINLAND
EDEN
I
__
y.
RUSSIA
RUSSIA
20 pm
POLAND
POLAND
v
700 rn 26.4.1986 03:00 UTC
700 m 26.4.1986 03:00 UTC
Fig. 1. Trajectories of an air parcel and particles 10 gm and 20 gm in aerodynamic diameter originating from Chernobyl on April 26, 1986 at 03:00 UTC (effective release height 700 m). Onthe left, convective uplift is not assumed, whereas air parcel and particles on the right experience uplift of 1000 m. Location of the updraft is shown as a cross near Chernobyl. On the left, trajectories of the particles are terminated due to sedimentation; on the right, particle 10 gm in aerodynamic diameter do not fall to the ground, because at long distances large-scale upward movement of air exceeds the sedimentation velocity (see Fig. 2).
414
Valkama and P611iinen 2000
Ukra~e
Belarus
! Lithuania
i Bakic Sea
I Sweden
i
Finland
Air ~ . . ~ _ . 1500
-l-O~m~]~ .....
1000
.
.
.
.
.
.
.
Air 500
10 ~tm 0
200
400
600
800
1000
1200
, 1400
1600
1800
Distance (km) along the trajectory
Fig. 2. Height of the trajectories presented in Fig. 1 as a function of distance. Thick lines represent air parcel and particles 10 ~tm and 20 ~tm in aerodynamic diameter that do not experience vertical uplift. Thin lines refer to air parcel and particles that experience updrafts of 1000 m in Ukraine.
CONCLUSIONS After the Chernobyl accident highly radioactive particles were found at longer distances than expected. Either the maximum effective release height was considerably larger than previously reported or convective currents of warm air lifted the particles upwards, to layers where different wind conditions prevail. During the early stages of the accident convective uplift may have transported radioactive materials from lower air layers, originally transported to Sweden, to higher air layers in which materials were transported to Finland. The radiological consequences might have been different had the accident occurred in the day time, when convective clouds would have been present near Chernobyl. The present long-range transport and dispersion models developed for radioactive substances are not necessarily adequate to describe the radiological hazards, as long as convective updraft in cumulonimbus clouds is not included in the models.
REFERENCES Bonelli, P., Calori, G. and Finzi, G. (1991) The influence of deep convection phenomena on trajectories computed by long-range transport models. Proceedings of the 19th International Technical Meeting on Air Pollution Modelling and its Applications, CCMS/NATO, Athens 1991. George, J.J. (1960) Weather forecasts for aeronautics. Academic Press, New York. Valkama, I., Salonoja M., Toivonen, H., Lahtinen, J. and P611~inenR. (1995) Transport of radioactive gases and particles from the Chernobyl accident : Comparison of environmental measurements and dispersion calculations. Proceedings of an International Symposium on Environmental Impact of Radioactive Releases, Vienna, 8-12 May 1995, IAEA-SM-339/69, 57-68. P611~en, R., Valkama, I. and Toivonen H. (1996) Transport of radioactive particles from the Chernobyl accident. Submitted for publication.
ON THE P H Y S I C A L , C H E M I C A L ~ A N D T O X I C HIGHLY DISPERSED ATMOSPHERIC K .R. S P U R N Y
Aerosol Chemist,Eichenweg 57392 Schmailenberg,Germany
PROPERTIES AEROSOLS 6
OF
,
Abstract- The importance and actuality of highly dispersed atmospheric aerosols(HDAA) had increased considerably during the last decade.There has been recognized that this fine fraction of aerodisper~ sed air pollutants is probably much more involved in the health ef fects of the general population than previously thought.The toxicity of HDAA can be correlated to their physical and chemical properties . Keywords
-Aerosols;Particle
Sizes;Particle
Chemistry;Toxicity
INTRODUCTION
Recent epidemiological studies indicate health effects on general population at air particulate concentrations laying below the existing air quality standards.They indicate that increases in human mortality and morbidity have been associated with levels of air particulate polutions significantly lower tha these previously thought to affect human health. In the majority of published papers,concentrations of the total aerosol(TSP-total suspended particulates)and/or the PM-10 fraction (air suspended particulate with an aerodynamic particle diameter less than i0 ~m)were correlated with the observed health effects.The de tailed chemical composition of the TSP or PM-10 was not considered.Nevertheless,this fact is a crucial point.The air particulate mass(TSP and PM-10)are as a matter of fact an unsuficiently defined and obscure air pollution standard.TSP and PM are not one pollutant,but are a class of pollutants.TSP or PM with no designation of their chemical and physical nature violates every principle of toxicology.THe consequences are needs for new definitions,better sampling and measurement methodology as well as for new air quality standards(AQS).
E P I D E M I O L O G Y AND T O X I C O L O G Y The mentioned epidemiological studies demonstarte positive correlation ons between particulate air pollution concentrations(mainly PM-10)and different health effects.These effects ~ere onserved at particulate mass concentrations as small as I0 ~g/m .The health effects were more aggravated for children,elderly and vulnerable sequences of p o p u l a t i ons(Pope et al. 1995,Withey 1989,Lipfert and Wyzga,1995).Examples of such results are schown in Fig.l. The i n v o l v i n s mechanisms of these adverse effects are nevertheless practically unknown.There is toxicological evidence(animal inhalation experiments)for adverse health effects from pulluted air.Numerous controlled toxicological investigations of individual chemical species have clearly shown that specific constituents of ambient air particulate matter are associated with adverse biological effects,including carcinogenicity.The atmospheric particulates are heterogeneous mixtures which present difficult problems for toxicological studies and risk assessment(Schlesinger,1995).
415
Spumy
416
ql
9 qlP
9
9 9
9 9
e
~176176'~
9
9
9
9
o 9
e]Po
~176
9
9
9
o~ ~
9
700
i,"o"6001
Fine par~cles, pg/m 3
5
15
20
30
25
35
Particulates and m o r t a l i t y
.~_
g
O-
~ w ~=
9
9
9
9
O"
9 9149 9
e ,~e - o e ' ~ Z, ~176 e 9 0~ I P ~ * o
9
9
.
9 e l q~ o.
O 9
~
~
9 90
e~ ~
9
sulfate concentrat0on, pg/m ~
0
5
10
15
20
25
Figure 1 Mortality rates plotted against mean fine particulate pollution(above)and against mean sulfate air pollution levels (Pope et al. 1995) AIR P A R T I C U L A T E
CHEMISTRY
The atmospheric anthropogenic aerosol(AAA)is an aerodispersed system of solid and liquid particles with different sizes,particle forms and
Physical, chemical, toxic properties of highly dispersed atmospheric aerosols
417
chemical composition.Only a few part of these particles,mainly the primary ones,are single inorganic or organic compounds.The majority of these particles reperesent chemical mixtures with complex health effects(Tab.l). Table 1 :Fine AAA and their health effects
Fine particles profile
Health effects of particulate matter
Formation processes: chemical reacuon, nucleation, condensation, coagulation, evaporatJon of fog and cloud droplets in which gases have dissolved and reacted Composition: sulfate, nitrate, ammonium, hydrogen ion, elemental carbon, organic compounds, PNA, Pb, Cd, V, Ni, Cu, Z.n, particle-bound water, biogenic organics Solubility:. largely soluble, hygroscopic, and deliquescent Sources: combustmn of coal. oil, gasoline, diesel, wood; atmospheric transformatmn products of NO~, SOv and organics including biogenic orgamcs such as terpenes; high-temperature processes, smelters, steel mills Ufetime: days
Reportedhealth effects associated with particulate matter exposures 9Mortality , Increased hospital use: admissions, emergency room visits 9Increased pneumonia and exacerbation of chronic obstructive pulmonary disease: hospital admissions, emergency room visits 9Exacerbations of asthma: attacks, broncho&lator use, emergency room visits, hospital admissions 9increased respiratory symptoms: cough, upper and lower respiratory tract problems 9Decreased lung function
The AAA are produced in the atmosphere and undergo during their residence time several physical,physico-chemical and chemical processes. Their resulting product is a polydisperse system of chemically heterogeneous particles with a complex toxic and carcinogenic potential . Biochemically active components may have been present at the particle surface or inside the particles(Spurny ,1993,Kao and Friedlander,1995~ The AAA are formed by two basic mechanisms:Dispersion and condensation,including chemical reactions.The total respirable AAA fraction is a mixture of the primary and secondary aerosol and will have a bimodal mass size distribution:as both components are of different origin,they have also different chemical composition(Fig. 2 ).The fine fraction of AAA (PM~i with particle sizes less than of about 2-3 ~m)is associated wit~ ~ e most toxic,inorganic and organic air pollutants.This AAA fraction could be also designat ed as the "Lung Toxic Mode Fraction" and could serve as a possible m arker for the dosage of toxic and carcinogenic components of partic ulate air pollutants.Smaller particles tend to have uniform spatial di stribution in the ambient air and the ability to penetrate into deep lunglth~ff~iPM c~se(Pr~ PMl0.PM2'5)sh~ be therefore a better index of hea
PMfi neMONITORING There do exist considerable var iations in the size distributions of ambient particles .However, there i s generally a clear separation into fine and coarse modes,with a dividing point between 1.0 and 2.5 um where the mas of the two modes is at a minimum.Condensation aerosols normally do not grow above 2 ~m and significant concentrartions dispersion aerosols are not normally found below 2 ~m(Lundereen and Burton 1994; Chow et a].,19q4). The necessary k n o w - h o w and equipments for the samDlin~ of PMf ereS~. PM is alreaSy available(Willeke and Baron, 1993)-Very comple~n-ch emica~'5nalysis of the AAA samples is now well possible and useful for several basic studies in the atmospheric envlronment and in the toxico-
418
Spumy
logical reserach.Nevertheless,practical applications of such results in the health risk assessment is difficult.Furthermore,the chemical analysis or identification of some special markers could be sufficient in routine measuremnets.Maybe,besides the total M~ mass,a continuous,in situ and on line real time monitoring of car~6~ black and PAH-aerosols could be a useful alternative solution. J
i
lll
i
I
I
I
I I I II I
I
I
10, Waldmann's analysis seems in very good agreement with experimental results. The intermediate regime (l
0,12 O.lO
5,6
0,08
c>... "-----" o,o6
o,8
I
x.,... s LL
1
v
._____-
O,IN
I
0,02
0,6
-
0,00 0,5
0,0
i
i
i
i
1
i
l
i
0,2
0.4
0,6
0,8
1,0
1,2
1,4
1,6
1
1,8 2,0
0.00
Kn -1
i
!
1
!
0,20
0,40
0,60
0.80
1.00
Kn
Fig. 3a Reduced thermophoretic velocity as a function of K n -1
Fig. 3b. Reduced thermophoretic force as a function of Kn
In Fig. l a, 2a and 3a we have represented the reduced thermophoretic velocity, vth/(rl/pT) V'I', as a function of Kn 1. The dependence of the thermophoretic velocity on the particle size is apparent. In Fig lb, 2b and 3b the experimental reduced thermophoretic force, Ft~/R 2 VT, data are compared with the known theories. It is clear that the Jacobsen' s equation (3) best interpolates our experimental values.
COMPARATIVE DATA ANAL YSIS In Fig. 4 the reduced thermophoretic velocity data of carnauba wax, PSL and silver together with those previously obtained by Prodi et al. (1979) with sodium chloride aerosol particles are represented. Considering the carnauba wax, PSL and NaC1 particles it appears that the thermal conductivity of the materials does not affect the behaviour of particles up to Kn = 4. For Kn ~ >4, in spite of the fact that we do not have experimental values over the entire range of the Knudsen number for each material, it
450
Prodi et al.
seems that particles with higher thermal conductivity assume higher thermophoretic velocity. The peculiar behaviour of silver can be explained considering the very high thermal conductivity of this material compared to that of the other particles. 1,1
O NaC1 experimental values 121 carnauba wax experimental values / x PSL experimental values silver experimental values
1,0 0,9 0,8 0,7 0,6
v
>-=2 0,5 I
o
o oA
0,4
~5
0,3 0,2 o,1 o,o 0
1
i
i
i
i
2
4
6
8
10
Kn-1 Fig. 4. Comparative reduced thermophoretic velocity as a function of Kn"l CONCLUSIONS From the comparison of the experimental data obtained with carnauba wax, PSL and silver aerosol particles with the known theories, it is clear that the Jacobsen's equation best interpolates our values. From the comparative data analysis, considering the NaC1, carnauba wax and PSL particles, the thermal conductivity seems to influence the thermophoretic velocity for Kn1 >4. The unusual behaviour of the silver particles can be explained considering the very high thermal conductivity of this material. Atter realising some more thermophoretic velocity measurements with very high thermal conductive metallic aerosols this research will be pursued performing scavenging experiments with these particles. Brock, J.R (1962) Jr. Colloid Sci. 17, 768. Derjaguin, B.V. and Bakanov, S.P.(1961)Dokl. Akad. Nauk. SSSR 141, 384 Derjaguin, B.V., Storozhilova, A.I. and Rabinovich, Ya.I. (1966)J. Colloid. Sci. 21, 35 Derjaguin, B.V. and Yalamov, Y.I. (1964) Dokl. Akad. Nauk. SSSR 155, 886 Epstein, P. (1927)Z. Physik 54, 537 Jacobsen, S. (1964) M. S. Thesis, Dept. of Chem Eng., Univ. Texas, Austin, Texas Li, W. and Davis, E.J. (1995) Jr. Aerosol Sci. 26, 1063 Prodi, F. and Oraltay, R.G. (1992) Precipitation scavenging and atmosphere-surface exchange, Vol 1, Hemisphere publishing corporation Prodi, F., Santachiara, G. and Prodi, V. (1979) Jr. Aerosol Sci. 10, 421 Prodi, F. and Tampieri, F. (1982) PAGEOPH 120, 286-325 Prodi, V. (1972) 'Assessment of Airborne Particles', edited by Mercer, T.T., Morrow, P.E. and St6ger, W., C.C. Thomas publ., Springfield, Illinois, 169 Rosenblatt, P and LaMer, K.H.(1946) Phys. Rev. 70, 385 Schmitt, K.H. (1959)Z. Naturforschg. 14a, 870 Tong, N.T. (1975) J. Colloid Interface Sci. 51, 143 Waldmann, L. (1959) Z. Naturforschg. 14a, 589
INTERNAL/EXTERNAL
MIXING
OF AEROSOLS
BY
COAGULATION
A. FASSI FIHRI, K. SUHRE and R. ROSSET Laboratoire d'A&ologie (UMR CNRS/UPS 5560) O.M.P, 14, Avenue Edouard Belin, 31400 Toulouse, France A b s t r a c t - A sectional model is used to simulate internal/external mixing of atmospheric aerosol particles by coagulation. The model initially consists in two externally mixed particle types made of pure substances A and B which collide to form a third population of internally mixed particles that contains all externally mixed components. For the tests, an exact solution is obtained that gives the total number concentration of each type of particles for a constant coagulation kernel. Good agreement is found between the predicted and the exact total number concentrations. The model is then applied to simulate a plume with mixing evolution of soot and hygroscopic particles. Optical and hygroscopic properties are calculated for all types of particles to bring out their sensitivity to particle composition, size and mixture type. K e y w o r d s - aerosol modeling, sectional model, aerosol mixing, radiative forcing, soot-sulphate aerosols INTRODUCTION Aerosol particles have a direct effect on climate through partial reflection of the incident solar radiation, thus affecting the solar heating distribution throughout the atmosphere. They also have an indirect effect as they may serve as cloud condensation nuclei (CCN) which affect cloud formation, their optical properties (albedo) and the precipitation processes. The direct and indirect radiative effects of aerosol particles depend on their size-distribution, their structure and their chemical composition which is quite diverse, reflecting their various sources and formation processes (homogeneous and heterogeneous nucleation) and evolution processes (coagulation, condensation, cloud cycling, deposition,... ). During the transport of atmospheric aerosols, mixing between aerosol populations of different source regions and composition takes place. Aerosol spectra resulting from these mixing processes have different size and composition which make their study difficult. This study is intended to show in a highly simplified way the effect of internal (IM) versus external (IEM) mixing on the optical and hygroscopic properties of an hybrid aerosol population evolving through coagulation and partial mixing of two initially chemically different populations. The sectional model of Warren and Seinfeld (1985) is extended to simulate the IEM aerosol issued from two aerosol populations by coagulation. The model is validated through comparison between an analytical solution for an idealized case of IEM aerosols, and through comparison with experimental data relevant to the mixing of a soot aerosol population locally emitted at an industrial site, with a background aerosol population made of soluble material (Okada, 1983; Strom et a/.,1992). The model is then applied to simulate the mixing of a plume of soot aerosol in the nucleation mode with a second aerosol population made of sulphate particles in the accumulation mode. Light scattering, absorption coefficients and the evolution of aerosol spectra due to varying relative humidity are calculated and compared for both the IM and IEM versions.
451
452
Fassi Fihri et al.
A- MODEL DESCRIPTION AND VALIDATION The model is an extension of the sectional IM model of Gelbard et al. (1980) and Warren and Seinfeld (1985) allowing for IEM between two different aerosol populations by Brownian, gravitational and turbulent coagulation effects. The prognostic variables are the total aerosol mass in each size bin for pure A, pure B, total M, and the mixing ratio A/M (from which B/M is deduced). The particle size domain is divised into 43 sections ranging in diameters between 0.001 #m and 25 #m. To validate the model, an analytic solution for the total number concentration of each particles type is calculated for a constant coagulation kernel/3. It is given by: Ni(t) N~(t)
= -
(2 +/~tN~ N(t)-
4N~ + l ~ t ( N ( t ) - N~
N~(t) -
gB(t);
i = A, B ,
(1)
2N o N(t) - 2 + ~tNo
where NA, NB and NM are respectively the total number concentrations of particle types A, B and M and N ~ the initial total number concentrations for particles of types A and B. N ( t ) represents the analytic solution for the total number concentration (Smoluchowski, 1917) and N o, the initial total number concentration. Tests with different synthetic distributions have been made and confronted to the solution given by Eqn. (1): they show that the maximum error in the total number concentration for all aerosol types (A, B, and M) after 10 hours of simulation is always below 6% for each individual aerosol chemical component, with practically no error in mass conservation due to the construction of the model itself. The model is tested again comparing its results to experimental data for a case described by Strom et al.(1992). The focus here is mainly on the simulation of the percentage of mixed particles as a function of particle size as observed by Okada (1983) in a rural area of Japan. The model results (not shown for brevity) are well within this uncertainty, but like the model of Strom et a/.(1992), it underestimates the percentage of measured mixed particles. B- AGING OF SOOT PARTICLES IN A PLUME The model presented and validated so far is now used to study the aging of a plume of soot particles submitted to mixing with a typical marine background air mass. Type A aerosols are from now identified with soot particles, whereas type B aerosols are assumed to be sulphate particles. The model is set up as a well mixed Lagrangian box model of volume V(t), in order to simulate the evolution of a moving plume cross section emitted at the source. The plume expansion is described by a constant horizontal diffusion (Fig. 1). Instant mixing is assumed between the plume and the entrained ambient air. The model is initialized with a monodisperse peak of 50000 #g m -3 primary soot particles of radius 0.01 #m and with 10% sulphate in the same size bin. The addition of sulphate to the initial soot distribution accounts for the fact that during the combustion process, sulphates are formed that eventually condense on primary soot particles or nucleate in order to form pure sulphate particles (Liousse, 1993). These high concentrations, which reflect the initial combustion process, are rapidly diluted with the background air. During the first hour, this peak is primarily shaped by Brownian coagulation to form a nucleation mode peak at 0.02-0.03 #m. Simultaneously, these nucleation mode particles coagulate with accumulation mode particles from the background atmosphere as they are entrained and mixed into the plume. As the plume is further diluted, nucleation mode particles are consumed by coagulation with accumulation mode particles. Most pure soot and sulphate particles from the background spectrum are transformed through this process into mixed particles. The contribution of the nucleation mode aerosols to the total extinction coefficient continuously decreases as the plume is diluted with the background air. Neither the mass distribution nor the
Internal~external mixing of aerosols by coagulation
453
contribution to the extinction coefficient for particles larger than 0.1 #m clearly change during the simulation, which is due to coagulation time-scales for this mode typically larger than one day. background mass
mass - ~ -
-~,m__i]
extinction
i
number ~ I
]t~JL._~
AA[
----
I
l
~
background number
~'-- -
I t=6h
--t=24h
Fig. 1" Geometry of plume mixing with background air. A Lagrangian cross sectional box of the plume is followed for t = l h , 6h and 24h after emission. The model is initialized with a monodisperse peak of 50000 #g m -3 primary soot particles of radius 0.01 #m and with 10% sulphate in the same size bin. The box volume grows and background air is entrained at a relative plume expansion velocity of V / V - 1/2t. Based on the scenario above, differences in optical and hygroscopic aerosol properties related to IM versus IEM modelling are now examined. Mie theory is used to calculate absorption, scattering and extinction coefficients. Hygroscopic properties are calculated using empirical growth laws (Sloane 1984; Lowenthal et al., 1995) and assuming a relative humidity increase from initially 50% to 90%. It can be observed in Figure 2-a that total light extinction is hardly sensitive to differences in IM and IEM modelling, whereas IEM is more diffusive and less absorbant (about 5% after one day of simulation) than IM in the present case. In order to study the hygroscopic growth properties of the aerosol spectrum obtained after 24 hours of simulation, the number of particles larger than a given critical radius r0 is calculated for both IM and IEM model types after application of hygroscopic growth. As a first approximation, particles larger than a given critical size are assumed to be subject to activation at a given Arsup /Arsup t h u s g i v e s a n e s t i m a t i o n supersaturation during the cloud formation process. The ratio I,IEM/~,IM for the relative difference in the number of CCN respectively available in the IEM and in the IM cases. As shown in Figure 2-b, the IEM model yields a higher number (up to 14%) of potential CCN than the IM model. The difference is most important for particle sizes that compare to typical smallest-activated-aerosol sizes during cumulus cloud activation (Ghan et al., 1993). C- CONCLUSION The climatic impact of aerosols results in a complex function of their size distribution, chemical composition and internal structure. From this standpoint, knowledge of the radiative and hygroscopic properties of atmospheric aerosols according to their state of internal/external mixing is still a pending problem. In this study, a sectional model has been adapted to investigate this problem in terms of order-of-magnitude arguments only, due to the simplifying assumptions made.
454
Fassi Fihri et al.
After some analytical and experimental tests, the model has been applied to a scenario of plume mixing with its ambient air, representing one special case of temporal evolution of the state of mixing of a typical aerosol population. 10 ~
5
- - - -
.= 10
o
=: 0
05
~-5 -10
'
.... - . . . . . 00
0
5
10 15 Timein hours (a)
20
0.1
1.0 radius in microns (b)
Fig. 2: (a) Evolution of the relative difference between IEM and IM(in %) for total light extinction (solid), absorption (dotted) and scattering (dashed) coefficients (b) Relative difference between IEM and IM in the number of particles larger than a given radius hfsu p ! Afsup "IEM/I'IM - 1 ) after hygroscopic growth for an increase in relative humidity from 50% to 90o-/6 . It has been found that in this particular scenario, namely modelling the aerosols as an internal/external mixture, rather than simply assuming internal mixture, yields some differences in the aerosol optical and hygroscopic properties. In the IEM case, the obtained aerosol spectrum is more light diffusive and less light absorbant than in the IM case, while more CCN can be activated from this spectrum. However, if the general computational framework can be kept, the particular results obtained have to be considered in the context of the approximations made in this scenario. Further scenarios and experimental tests are in order to assess the impact of internal/external mixing on aerosol properties and to put this in the context of global aerosol forcing modelling. REFERENCES Gelbard F., Tambour Y. and Seinfeld J. (1980) Sectional representations for simulating aerosol dynamics. J. Colloid Interface Sci. 76,541-556. Ghan S. J., Chuang C. C. and Penner J. E. (1993) A parameterization of cloud droplet nucleation. Part 1. single aerosol type. J. Atmos. Res. 30, 198-221. Liousse C. (1993) Emissions carbondes particulaires des feux de savane d'Afrique: mesures au sol et tdldddtection spatiale des panaches. PhD thesis, University of Paris VII, France. Lowenthal D. H., Rogers C. F., Saxena P., Watson J. G. and Chow J. C. (1995) Sensitivity of estimated fight extinction coefficients to model assumptions and measurement errors. Atmospheric Environment 29, 751-766. Okada K. (1983) Nature of individual hygroscopic particles in the urban atmosphere. J. Meteor. Soc. Japan 61,727-736. Sloane C. S. (1984) Optical properties of aerosols of mixed composition. Atmospheric Environment 18, 871-878. Smoluchowski M. V. (1917) Versuch einer mathematischen Theorie der Koagulationskinetik kolloidaler LSsungen. Z. Phys. Chem. 92, 129-168. StrSm J., Okada K. and Heintzenberg J. (1992) On the state of mixing of particles due to brownian coagulation. J. Aerosol Sci. 23, 467-480. Warren D. R. and Seinfeld J. H. (1985) Simulation of aerosol size distribution evolution in systems with simultaneous nucleation, condensation and coagulation. Aerosol Sci. Technol. 4, 31-43.
NUMERICAL
MODELLING AND
OF
AEROSOLS
IN
MINOR THE
GAS
CONSTITUENTS
ATMOSPHERE
A.E. A L O Y A N Institute of Numerical Mathematics, RAS, 32a Leninskiy Prospect, Moscow, Russia A b s t r a c t - Consideration is given to the simultaneous model of mesoscale atmospheric hydrodynamics and pollutants transfer taking into account their transormations. Namely, photochemical and chemical reactions proceeding in liquid and gaseous phases as well as atmospheric aerosol transfer considering condensation and coagulation. The results of numerical experiments are given. K e y w o r d s - Aerosol transport; Turbulence; Photochemistry; Coagulation; Condensation. INTRODUCTION A great deal of gaseous and aerosol-state chemical substances are blown out into atmosphere in consequence of human economic activity. These substances are subject to a series of physical and chemical variations by photochemical transformation, coagulation and condensation mechanisms. All these mechanisms are interdependent as parts of the general complex ecological problem. Sulphur and nitrogen oxides, metallic dust and other substances are of particular hazard among antropogenic effluents. For example, sulphur oxide effluents assist in acid rain precipitation when turning into aerosol phase. In turn, that has a severe impact on the Biosphere. Besides, acid rain gives rise to some toxic effects observed in smog. In this case it is of high importance to clarify gas-to-particle conversion mechanisms which give rise to substances capable of condensation, such as sulp'huric acid, ammonium sulphate, ammonium nitrate etc. Let us now consider the ecological problems characterized by necessity to describe photochemical transformation mechanisms of aerosol formation processes by coagulation and condensation, transport of conservative substances etc. The circle of these problems is rather wide so we shall fix our attention only on some of them. Note that all these problems are solved in conjunction with atmospheric hydrothermodynamics models, but here we restrict our consideration to the equation of pollutant transport only. Represent this equation in generalcase as follows 0--~-t- V(u,~i) - Fi + V ( k V ~ i ) + B i j ( x , ~ i , ~ j ) + / ~ / ( x , ~ g , ~g_g~)+/Sij(~g, ~ g , ) + I(c2g, t),
(1)
where k is the diffusion coefficient, x is the radius-vector, Fi are effluent sources, Bij,/1),/Sj are non-linear operators describing photochemical transformation, coagulation and condensation respectively. The boundary conditions are
~,lt=0 = ~o,
OcPi Oz
~=•
- O,
~
y=+y
= O,
Oz
= ao(9~, - r
(2)
where ao describes the pollutant interaction with underlying surface; ~i0 is the pollutant concentration when z = z~ and z~ is the roughness parameter. The kinetic equation being solved making use of the values of meteorological and turbulent characteristics of the atmosphere obtained from dynamic model.
455
Aloyan
456
CONDENSATION MODEL
The model of saturated vapors condensation permits to study the processes of fluctuative nucleation and further growing of the particles in the supersaturated vapor, which leads to the initiation and progression of the disperse phase. Consider that only the drops of size 7g* (where 7 > 1, g. is the number of molecules in a critical drop) are produced, and intensity J as well as critical size are varying with time. Then, according to Aloyan (1992), the kinetic equation in spatially-uniform case takes the following appearance:
Ofg ~Ot
= J ( t ) 6 [ g - "/g.(t)],
Og
(3)
where fg(t) is non-equilibrium distribution function of the particles spectrum, vg - ~Or" The model discussed permits to reproduce the bulk condensation processes in different situations, provided that the laws according to which the thermodynamic variables change are given. The processes of the sulphuric acid aerosol formation in the atmosphere are treated. Fig.1 presents the synthesized plots of the results obtained for the height z = 150m at the time moment t = 6p.m. At the left side of this figure the fields of vaporous acid concentration and velocity are presented, while placed at the right side are the logarithmic-scaled size distributions for the particles in liquid phase, at four various distances from the source.
Y,km]
t= 6pro.
z = 150m
....... ....ci', ::;;:::::1 'I--
........... . . . . . .
t
t . . .
~\~::..:.../ t,
r
~
,
i i :::::-:::::-
-;..,... o !!!!?iiii i::!}!i::
........... ; ; :; ;:
~
~
o,
-I,
~
". ' 'Ill ~ . . . . . . ] - " 0.0 : : : ; '0 : ~ ~ f : : ; / [r]--~um
: ~o~
~Nj=~m.|
l-i 0.4
-1.2 -0.8 -0.4
Logr
[r]=~zm
0.0
0.4
Logr
{
........
::.~~':::/-~t::l
:!..... I I I I
. .. ... ... .. ... ... ... ... ... .. . .."... ". ~"-"'~'~ . . . . . [-5 / - 4' . ,]. !~. - I:. . . . t , .I. J.I ,.!.. t ,.177~. I "I " : : ....
: . . . . . . . . . . . . . .
t
-1.2-0.8-0.4
............. ' ..... / [r]=om 6 . . . . I ~ "" 20() "" 300""" "4(1()'"" "500"x, km [ ] U~--2.5m/s,
0.0
0.4
-1.2-0.8-0.4
Logr
[r]=um
0.0 0 . 4
Logr
+Nm~ --5.7.10nmol/cm :~
Figure 1: Sulphuric acid aerosol in the atmosphere.
COAGULATION MODEL
Now turn to formulation of the kinetic equations of coagulation. A system will be considered to sufficiently sectional and the merging to take place only in pair collisions. Then the concentrations and particles size variation are described by Smoluchowsky equation (Zagaynov, Lushnikov (1988), Aloyan et. al. (1993)):
Ot -- -2
~'(g'g1)9%-g'~g'dgl 0
--
I('(g, gl)~g, dgl + I(cgg, t), ~glt=0 - 9 ~
(tog
(4)
0
where gl is the integration variable of the current mass; ~?g is the concentration of particles of mass interval [g, g + dg]; 1(c29, t) is the intensity of the formation of new particles of mass g. The
457
Numerical modelling of minor gas constituents and aerosols
first term in the right side of (4) describes the particle ingoing to the size g by minor particles coagulation. The second term describes the outgoing from the size g by coagulation with all other particles. The coagulation kernels are taken close to reality, which allows us to take into consideration several collision regimes (free molecular, transitions and diffuse ones). PHOTOCHEMICAL
TRANSFORMATIONS
The model under consideration allows to compute the transformations of sulphur and nitrogen containing species and methane by the photochemical mechanism. The numerical experiment of transformation modelling is follows: it is supposed that several trace gases are blown out into some model atmosphere of a city S02, NO, NO2, CO, CH4, CH20. Some of them (e.g. S02, NO2, CH20 turn to the excitation state selectively absorbing solar radiation in ultra-violet range. This promotes fast photochemical transformations that result in a series of chemically active atoms, free radicals etc. The photochemical model contains 170 reactions of 32 gaseous components. The left side of Fig. 2 shows in logarithmic scale the sensitivity of NO2, OH, H202, HN03, HN02, 03, HCOOH concentrations to variations of [NO• in range 109 - 10 TM. Time dependence for H + concentration in H202 and 02 oxidation, and oxidation catalysis with Fe and Mn ions has also been studied. It is obtained that in polluted atmosphere the latter channel is the major one. Fe and Mn ions serve as catalysts in the limiting stage of the sulphur dioxide oxidation chain mechanism. For the background atmosphere a strong dependence of the oxygen channel of S02 oxidation is revealed. The results for [H+] are illustrated at the right of Fig. 2. lgIN}
11.0 T x x x x x x x x ~ x x.x,X-x
g.O
%
.....
2~
~*~
2.00
,-~ 1.00
'',
4-
T ,--..a
" .....
/ ~)
0.00 l 0.00 g oo
~,oo
l l l U l l l l l l l l l l l l l l
~3oo
500.00
t, C
I
1000.00
Figure 2: Some results of the photochemical model. REFERENCES
A.E.Aloyan (1992) A non-hydrostatic numerical model of mesoscale atmospheric layer. Russ. J. Num. Anal. Math. Modell., v.?, No.5, pp.371-456. A.E.Aloyan, A.A.Lushnikov, S.V.Makarenko, G.I.Marchuk, V.A.Zagainov (1993) - Mathematical modelling of the atmospheric aerosol transfer with coagulation taken into account. Russ. J. Num. Anal. Math. Modell. v.8, No.1. A.E.Aloyan, G.I.Marchuk, V.A.Egorov and V.N.Piskunov (1992) Aerosol formation mathematical modelling with consideration for condensation kinetics. Russ. J. Num. Anal. Math. Modell. ,v.7, No.?, pp.457-471. A.E.Aloyan, V.O.Arutyunyan, G.I.Marchuk (1995) Dynamics of mesoscale boundary atmospheric layer and impurity spreading with the photochemical transformation allowed for. Russ. J. Num. Anal. Math. Modell., v.10, No.2, pp. '93-114. V.A.Zagaynov, A.A.Lushnikov (1988) Modelling of the atmospheric coagulating aerosol Atmospheric Aerosol and Nucleation/Ed. G.Valt and P. Wagner.-Springer Verlag. pp.93-95.
M 3 A MULTI
MODAL
MODEL
FOR AEROSOL
DYNAMICS
JULIAN WILSON AND FRANK RAES Environment Institute, European Commission Joint Research Centre, I-21020, Ispra (Va), Italy. Abstract - A simple model of aerosol dynamics suitable for use in 3-D eulerian atmospheric transport models is described, and its' performance by comparison with a full sectional model is briefly demonstrated. The potential for improving the performance of the model through the determination and use of additional regression parameters to obtain a "best fit" with the full sectional model is also demonstrated. Keywords - HE0 - HESO4 aerosol, modal model, regression fitting. INTRODUCTION M 3, is a simple model describing the physical processes controlling the evolution of atmospheric aerosols; nucleation, condensation, coagulation, cloud processing and removal. It is designed for use in 3-dimensional eulerian atmospheric chemistry and transport models which use operator splitting. The aerosol population is represented by a number of log-normal modes with prescribed geometric standard deviations of 1.4, but evolving geometric mean diameters. By prescribing a log-normal distribution and the geometric standard deviation, only the number and mass of each mode need be transported. A decoupled scheme of explicit semi-analytical solutions describes the concentration of gas phase precursor species, and the number and mass of each mode. An inherent problem of simple models of non-linear systems such as that of aerosol dynamics is that they tend to linearize the more important features of the complex models (Raes and Van Dingenen, 1995). In M 3 this potential limitation is reduced by use of a novel method of model fitting (Saltelli, 1995). The differences between M 3 and the full sectional model AERO2 (Raes et al., 1992), are minimised by identifying the regression fitting parameters that give the "best fit" between the two models. In the following, the model and the fitting technique are briefly described, and key results of a model fitting experiment illustrated. THE MODEL The version of the model described simulates the evolution of an H20 - H 2 S O 4 aerosol population, in the absence of clouds, using three modes. The change in the gas phase concentration, of HzSO 4 is considered in two parts in order to derive exact analytical solutions: & ~v2
,,I,
(1)
N,,
and
= 5tomtsol
-(y_.,.,..,v,c,+.
v',,
x' tn?o:l
where k I = reaction constant, tt and fl = nucleation parameters, N c = critical number of H 2 S O 4 molecules per nucleation cluster, N i = number aerosols in mode i, Ci = surface median condensation
458
M 3 a multi modal model f o r aerosol dynamics
459
coefficient for mode i, Vag the dry deposition velocity, 2.g the wet deposition velocity of H2SO 4 gas, is the average surface of mode i aerosols, and YI Y2 and Y3 are nucleation rate regression parameters. The nucleation term in (1) is modified from that of Jaecker-Voirol and Mirabel (1988) by the addition of a divisor which is a function of the existing aerosol surface, in order to simulate the "quenching" effect of existing aerosol on the nucleation rate, which is otherwise lost by splitting the equation into two. The resulting integral nucleation rate is used as the source in the equation describing the number of aerosols in the first mode. The general form of this equation is given below: dN dt
' -
1KN2- (KM 2 ""
+ Va'+ ~.~N + S
(3)
i
where Si is the formation rate of mode i aerosol, ksi is the self-coagulation coefficient for mode i, and kmu the coefficient for multi-modal coagulation between modes i and j. Explicit analytical solutions of (3) are used to describe the change in the number of aerosols in each mode. Thus, for the first mode, S 1 is the integral of (2) over At, while for the higher modes, S i is the integral of the self-coagulation term in (3) for mode i-I, again over At, to give a linear source. The change in the mass of aerosols in each mode is derived from (3), assuming that for selfcoagulation mode i looses mass and number, and mode i+l gains mass and number, while multi-modal coagulation causes a loss of mass and number from mode i, but a gain of mass only for mode j. A useful feature of the log-normal distribution is that given the count median diameter (CMD), and the geometric standard deviation (og), any other average diameter can be derived using the HatchChoate (1929) conversion equations. The radius of average mass for a mode is determined from the mass and number of aerosol, and from this the CMD. In the current version of M 3, the radii of average surface are used to calculate the condensation coefficients, and the radii of average volume used to calculate the coagulation coefficients. Condensation Condensational growth of the aerosols occurs by collision of HzSO 4 molecules in the gas phase with the aerosols. The condensation coefficient in (2) is determined from Fuchs (1934) for an aerosol with the radius of average surface, rn: 4rcDriy i C: I
4D avr
si
(4)
r
+ r
si
+ A
where D, v and A are respectively the diffusion coefficient, thermal velocity and mean free path length of an H / S O 4 molecule, 0t is the accommodation coefficient and is taken to be 0.3 (Raes and Van Dingenen, 1992), and ?'si is the condensation rate regression parameter for mode i. Unlike sectional models, where condensational growth is represented by moving aerosol from one size class to the next, in M 3 condensational growth increases the CMD of the aerosol mode. Growth of aerosol from one mode to the next is represented by the transfer of aerosols in mode i, that have diameters within 2),cOi of the CMD of mode i+l, to mode i+l, where )'c is a regression parameter. This is treated separately to (3). Coagulation The coagulation coefficients for collisions between both aerosols of the same mode and aerosols from different modes in (3) are determined from Fuchs (1964):
Wilson and Raes
460
16nDr y K
= G(r .r ) ij
"~
-4D
(5)
r +
v-7v
r + A/ v
where/), v, A/and G(r~ r~) are respectively the diffusion coefficient, thermal velocity, mean_free path length, and van der Waals factor (Van Dingenen and Raes, 1990) for an aerosol with radius rv = Yj (r, + rv.Q / 2, and ~ is a coagulation rate regression parameter. For serf-coagulation (Ks) i and j are equal, while for multi-modal coagulation (K,,,)j > i. FITI'ING M 3 TO AERO2 In the regression fitting experiments both M 3 and AERO2 simulate the evolution of an HzO HzSO 4 aerosol population for a sample of representative marine boundary layer conditions. Monthly average input parameters (temperature, relative humidity, pressure, cloud fraction, daytime average sulfate formation rate, average gas phase sulfate concentration, day-length, background aerosol type (D'Almeida et al., 1991), and gross transport rate from the grid box) were extracted for 64 randomly selected surface and 950-850 mbar level grid boxes from the sulfur version of the eulerian tracer transport model MOGUNTIA (Langner and Rodhe, 1991, Zimmerman et al., 1989). Boxes were assumed to be representative of marine conditions if all the surface underlying both the box and all the adjacent boxes did not include any land. The background aerosol type, cloud fraction and monthly average sulfate concentration were used to infer an initial particle size distribution, which was also assumed to be present in the adjacent boxes. Transport is thus simulated by depleting the evolving aerosol size distribution, and adding to it from the initial distribution, in proportion to the gross transport rate out of the selected grid box taken from MOGUNTIA. Briefly, the regression fitting experiments involve a search among the population of regression parameters (the y's in the above equations), using a genetic algorithm, for the combination of y's which minimises the differences between the results from the two models. The difference between the results of the two models is def'med by the sum of the squares of the differences function(SSD). The result of any experiment is therefore dependent upon how the SSD function is formed from what output parameters. Possible output parameters can include numbers and diameters of the aerosol modes, properties of the cumulative distributions calculated by the two models, as well as integral properties of the aerosol populations, such as mass, number or volume. The most important criterion is that the SSD function should avoid as far as is possible, comparison of artifacts of one model or the other. In the experiment presented here, the AERO2 distributions were converted into three modes and log-normal distributions fitted to them. The SSD function was defined as the sums of the squares of the differences between the numbers in each mode, and the differences between the CMDs of each mode predicted by the two models, summed over a five day period once the AERO2 simulation had reached a steady state of oscillation. Figures l a and b show the mean number and CMD of the third mode, over days 17- 26, as predicted by the two models, with all regression parameters set to unity, while figures 2a and b. show the same results but using regression parameters identified by a fitting experiment, which reduces the SSD from 4E14 to 1E5. In converting the AERO2 distributions into the three modes, the third mode was defined as aerosols larger than the critical diameter for activation at 0.2% supersaturation. The results show a marked improvement in the ability of M 3 to reproduce the AERO2 results, albeit at the expense of the agreement for the first two modes, which are not shown here.
M 3 a multi modal model for aerosol dynamics 106 t-c)
461
~, 1.oo E
105 7 E
oE 104 Z 9 103
0.10
r
(..) z o r
102 101 101
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
102
l
. . . . . . . .
|
. . . . . . . .
103 104 M~N mode 3
i
O.Ol O.Ol
. . . . . . .
105
106
.
.
.
.
.
.
.
.
i
.
.
.
. . . . .
O.lO M~CMD mode 3 (um)
1.oo
Figures la & b: Comparison of AERO2 and M 3 mode 3 aerosol mean number and diameter for days 17- 26; 64 cases; standard model. 106
1.00 E
105 ~D
o 104 z 9 103
E
0.10
r
r..) Z o t--,I
102 101 101
.
.
.
.
.
.
.
.
.
.
.
.
102
.
.
.
I
. . . . . . . .
I
. . . . . . . .
103 104 M3N mode 3
I
105
. . . . . . .
106
O.Ol O.Ol
.
.
.
.
.
.
.
.
|
.
.
O.lO M3CMD mode 3 (urn)
. . . . .
1.oo
Figures 2a & b: Comparison of AERO2 and M 3 mode 3 aerosol mean number and diameter for days 17- 26, 64 cases; "fitted" model. ACKNOWLEDGEMENTS We would like to acknowledge the invaluable contributions to this work of A. Saltelli. This work is undertaken as part of the EC Environment and Climate research project SINDICATE. REFERENCES D'Almeida G.A., P. Kopke and E.P. Shettle (1991) Atmopheric aerosols global climatology and radiative characteristics. Deepak Publishing, Hampton V A. Fuchs, N. (1934) Phys. Z. Sowjet 6, 225. Fuchs, N. (1964) The Mechanics of Aerosols. Pergamon Press, Oxford. Hatch, T. and S.P. Choate (1929) J. Franklin Inst. 207, 369. Jaecker-Voirol and P. Mirabel (1988), J. Phys. Chem. 92, 3518. Raes, F., A. Saltelli and R. Van Dingenen (1992) J. Aerosol Sci. 23, 759. Raes, F. and R. Van Dingenen (1992) J. Geophys. Res. 97, 12901. Raes, F. and R. Van Dingenen (1995) J. Geophys. Res. 100, 14355. Saltelli, A. (1995) personal communication. Van Dingenen R. and F. Raes (1990) J. Aerosol Sci. 21, Suppl. 237. Zimmerman, P.H., J. Feichter, H.K. Rath, P.J. Crutzen, and W. Weiss (1989) Atmos. Environ. 23, 25.
PHASE PARTITIONING
OF AEROSOL PARTICLES
DURING
CRYSTALLIZATION Z. GE, A. S. WEXLER, M. V. JOHNSTON Department of Mechanical Engineering, and Chemistry and Biochemistry, University of Delaware, Newark, DE 19716, U. S. A.
Abstract- Thermodynamic analysis has shown that phase partitioning of multicomponent aerosol particles occurs during crystallization. Particles dried from multicomponent aqueous aerosols do not have a homogeneous chemical morphology except at the eutonic point. Rapid Single-particle Mass Spectrometry (RSMS) has been used to explore the chemical composition of particles dried from KC1/NaC1, KC1/KI and (NH4)zSOn/NH4NO3 mixed solutions at different mole ratios. The results suggest that the surface layer is enriched with the minor component and are consistent with our thermodynamic predictions. Keywords- Atmospheric aerosols; Phase partitioning; Crystallization; Laser desorption/ionization. INTRODUCTION The mechanisms of crystallization of multicomponent aerosol particles are important in understanding the atmospheric processes affecting air quality, visibility degradation and climate change. Wexler and Seinfeld (1991) pointed out that the crystallization of multicomponent aerosol is dependent on the mole fraction of the compounds and the mutual deliquescence relative humidity. Consider an aqueous aerosol particle containing two salts, as the relative humidity is lowered, water evaporates from the particle to maintain equilibrium. Eventually one of the salts becomes saturated and forms its crystalline phase. As the relative humidity continues to decrease, more of this salt crystallizes and simultaneously the solution becomes more concentrated in the other salt. Eventually the other salt reaches its saturation. At the mutual deliquescence relative humidity, the two salts effloresce completely and a mixture of solid phases forms. To minimize the surface free energy, the first solid precipitate should be located in the center of the particle surrounded by the remaining salts. Thus, the dried particles are composed of a pure salt core surrounded by a mixed salt coating, where the core is solely determined by the original aerosol composition but the coating is identical in chemical composition to the eutonic point and is independent of the original aerosol mole fraction. Recently, a new technique(Rapid Single Particle Mass Spectrometry, RSMS) has been used to probe the chemical composition of particle surface (Carson et al., 1995). Monodisperse aerosols were produced from a vibrating orifice aerosol generator and transferred to the source region of a time-of-
462
Phase partitioning of aerosol particles during crystallization
463
flight mass spectrometer through a differentially pumped inlet. The particles were detected by light scattering with a continuous laser beam and the scatter pulse triggered an excimer laser which ablated each particle in flight. The resulting ions were accelerated into the mass spectrometer and detected with a dual microchannel plate detector. The output signal was sampled with a transient digitizer mounted in a personal computer, and the spectrum of each particle was recorded. By measuring the peak area ratios corresponding to the intensity of certain ions in the spectrum, the chemical composition of the surface layer can be inferred. In this study, three groups of particles dried from KC1/NaC1, KC1/KI and (NH4)2SO4fNH4NO3 mixed solutions at different mole ratios have been investigated. The summary of experimental systems is shown in Table 1.
Table 1. Summary of Experimental Systems
peak area ratio
mean particle diameter(lam)
K+(39), Na+(23)
[K+]/[Na +]
3.5
C1-(35, 37), KI2-(293)
[C1-]/[KI2-]
3.0
analyte ions in
quantitation ions
solution
(m/z)
KCI+NaC1
K +, Na +
KCI+KI
CI-, I-
(NH4)2SO4+
8042-, NO3-
salts
NH4NO3
HSO4-(97), NO2-(46), [HSO4-]/([NO2]+ NO3-(62)
3.5
[NO3-])
RESULTS & CONCLUSION The results show good consistency with the thermodynamic predictions. We observed that the surface chemical composition of particles was generally identical with respect to different original solution mole ratios. Also, the results suggest that the surface layer is enriched with the minor component. The results from KCI/NaC1 aerosols imply that the eutonic point for this mixture is at mKCl / mNacl=3/7 which is consistent with the findings in Tang et al. (1976, 1978). We also observed that the surface of particles dried from (Nt-In)2SOg/NHnNO3 mixed solutions contains substantial (NH4)2804 which may inhibit the evaporation of volatile NH4NO3.
464
Ge et al.
REFERENCE Carson, P.G., Neubauer, K.R., Johnston, M.V. and Wexler, A.S., J. Aerosol Sci. 26, 535 (1995). Tang, I.N., J. Aerosol Sci. 7, 361 (1976). Tang, I.N., Munkelwitz, H.R. and Davis, J.G., J. Aerosol Sci. 9, 505 (1978). Wexler, A.S. and Seinfeld, J.H., Atmos. Environ. 25A, 2731 (1991).
HETEROGENEOUS CONDENSATION PROPERTIES OF ULTRAFINE SOOT P A R T I C L E S (20 nm < Dp < 100 nm) A F T E R R E A C T I O N W I T H O Z O N E . R. KOTZICK, U. PANNE, R. NIESSNER Institute of Hydrochemistry, Technical University of Munich, Marchioninistrasse 17, D-81377 Mfinchen, Germany Abstract - Condensation properties of ultrafine artificial soot particles in the Aitken range (particle diameter between 20 nm and 100 nm) were investigated by means of a variable supersaturation condensation nucleus counter (VSCNC). Critical supersaturations have been determined for dry, monodisperse soot particles before, and after the reaction with ozone. For all particle diameters, a significant shift towards a more hydrophilic behavior has been observed. For a quantitative description of the condensation properties, the results have been compared to the Fletcher-theory of heterogeneous condensation on insoluble particles. Keywords- Soot, Ozone, Condensation-properties, Contact-angle INTRODUCTION As rain-out is by far the most relevant mechanism for particle deposition, the atmospheric residence times of ultrafine particles are mainly determined by their condensation properties. The critical supersaturation for particle activation is strongly affected by the composition, the structure and the surface properties of an individual particle. A composition dependent particle activation due to a soluble fraction present in carbonaceous aerosols was already the subject of several investigations (Hagen et aL, 1989; Hailer et al., 1989). Also, the influence of the structure on the condensation properties has been widely studied (Colbeck et al., 1990), and theoretically calculated (Crouzet and Marlow, 1995). Contrary to these investigations, studies on aerosols to reveal the influence of the surface properties are hardly available. An important atmospheric process affecting particle surface properties is the interaction between carbon aerosols and ozone. It is now commonly accepted, that soot can act as sink for atmospheric ozone as reported (Stephens et aL, 1986), resulting in the formation of gaseous reaction products like 02, CO and CO2. An experimental approach to study the carbo/a-ozone interaction is to use an in situ technique, avoiding artifacts from structural changes and soluble fractions. The objective of this work was to investigate the condensation properties of carbon aerosols in the Aitken range and changes in the hygroscopicity due to the reaction with ozone. Evidence will be given of the strong influence of the chemical surface properties on the activation of soot particles in the condensation process. For these experiments, an in-situ technique was employed, using a variable supersaturation condensation nucleus counter coveting the range from 10 % up to over 200 % supersaturation. For the purpose of this study, the Kelvin-equivalent diameter of an aerosol particle is defined as the diameter of a water droplet which starts to grow at the same supersaturation as the particle under study. The theory for the condensation on insoluble particles used for the description of the investigated aerosols, was introduced by Fletcher (1958) proposing the contact-angle | as measure for the hygroscopicity of an aerosol particle.
465
Kotzick et al.
466
EXPERIMENTAL For all experiments, pure carbon aerosols from a commercial soot generator based on a spark discharge between two carbon electrodes were used (Helsper et al., 1993). Monodisperse fractions have been selected with a differential mobility analyzer (DMA). This particular aerosol shows a fractal dimension of approximately Df = 2, similar to urban aerosols. Particle distributions were characterized with a commercial differential mobility particle sizer (DMPS). Monodisperse particle concentrations have been produced in the range from 103-105 particles cm -3, counted with a condensation nucleus counter (CNC). Ozone was generated by photodissoziation of 02 using UV-radiation from a Hg-line source in a simple flow generator. With additional dilution and mixing, the O3-concentration was varied between tropospheric (clean areas: 20 ppb - 80 ppb; polluted air: 100 ppb - 500 ppb) and stratospheric (up to 8 ppm) relevant concentrations. The 03 concentration was measured with a commercial O3-analyzer based on the UV-absorption of ozone at 250 nm. All experiments were performed in a laminar flow reactor, consisting of a glass cylinder (12 cm inner diameter, 140 cm length) with conical endings and axial sampling probes placed along the central axis of the flow reactor at 70 cm and 140 cm. Assuming a parabolic flow, the interaction times were calculated as 1 s, 60 s and 120 s for the three sampling points. Mixing of the aerosol with the ozone was achieved in a ring-gap mixer directly mounted to the reactor inlet. For complete removal of the ozone, an activated carbon scrubber was placed at the outlet of the flow reactor. The complete setup is shown below in figure 1. N V FM HV RGMN UV O3A VSCNC CNC ACS
Neutralizer Valve Flow Meter High Voltage Supply Ring Gap Mixing Nozzle 03 Reactor O3Analyzer VariableSupersaturation Condensation Nucleus Counter Condensation Nucleus Counter Activated Carbon Scrubber
FM V F ~ZZSq ~.T..1 ~---~. N2
iiii:................) Soot Aerosol Generator
i'. ,
.':~
Differential Mobility Analyzer
RGMN --
Figure 1"
HV '
FM
IV
i
v
Laminar Flow
t1
N~ = ~ - j
Experimental Setup.
The critical activation for the ultrafine particles was determined with a modified discontinuous NolanPollak counter. In this system, the particles are activated in an expansion chamber and the concentration of dry condensation nuclei at normal pressure and the absolute number of particles activated at a certain supersaturation are registered with two commercial particle counters of different type. To determine the initial dry particle concentration the particles are sampled with a condensation nucleus counter (CNC)
Heterogeneous condensation properties of ultrafine soot particles
467
(,TSI Model 3020), whereas the activated droplets are introduced into an aerodynamic particle sizer (APS) (TSI Model 3010). Overall, the variable supersaturation condensation nucleus counter allows the reliable generation of supersaturations between 10 % and 200 %, which enable us to determine the critical activation of strong hydrophobic carbon particles down to 10 nm particle diameter. The verification of the concentration measurements for the original and activated particles was performed with dry, monodisperse carbon particles and revealed a good linear relation between different initial particle concentrations and the number of activated particles registered by the APS. RESULTS First experiments using the artificial ultrapure carbon aerosol (no soluble, e.g., sulfuric acid fraction) were made under various experimental conditions, e.g., ozone concentration (20 ppb - 1 ppm), reaction times (1 s to 120 s), and particle diameters (20 nm - 100 nm), respectively. Ozone treatment resulted in a significant shift of the Kelvin-equivalent diameter of carbon aerosols to lower critical supersaturations for particle activation. This was found for all ozone concentrations and different particle diameters. The relative magnitude of this shift was larger for smaller particles (< 30 nm) than for particles with diameters > 50 nm. As the condensation characteristics of small particles are dominated by the spherical primary particle structure, the relative shift is expected to be larger than for particles with an extended agglomerate structure facilitating condensation also at structural boundaries. Within the experimental error, the same critical supersaturation for different ozone concentrations was observed for all reaction times and particle diameters, even at the lowest O3-concentration (20 ppb). 0.01 100
0.1
I
'
'
'
'
'
'
'
'
I
100
!
10
r~ o
[ ~
-
10
....
.
.
.
.
.
.
.
.
0.01
0.1
Dry Particle Diameter D [lam] P
Figure 2"
Critical supersaturations of ozone treated and untreated carbon aerosols. diamonds:
C particles
n = 3
uptriangele:
C particles
n = 3,
c . . . . = 1 p p m , ge.r
open uptriangle
C particles
n = 3,
c . . . . = 0.25 ppb, t~eoc~o. = 120 s
d o t t e d line:
Fletcher-theory, |
= 26 ~ P r u p p a c h e r and Klett ( 1 9 7 8 )
dot d a s h line:
Fletcher-theory, |
= 50 ~ P r u p p a c h e r and Klett ( 1 9 7 8 )
= 120 s
468
Kotzick et al.
This implies that all reactive sites on the surface of fresh soot particles could react with ozone during the studied interaction time. A comparison between measured critical supersaturations of untreated and ozone treated carbon aerosols with the Fletcher-theory for heterogeneous condensation on insoluble particles is given in figure 2. Based on the definition of the critical supersaturation, as the point when 50% of the particles are activated, the results from figure 2 show, that the contact angle of the small (< 30 nm), untreated carbon particles approaches -50 ~ whereas the deviations for increasing particle size can no longer be described by the Fletcher-theory due to the shape and structure of the agglomerates, which are violating the underlying assumption of spherical particles. On the other hand it can be clearly seen, that the ozone treated carbon aerosols can be described with a common contact angle of--26 ~ for all particle sizes. Obviously, the change of the surface properties dominates the condensation properties of the particles, while the agglomerated particle structure is now of minor importance. Results from a FTIR investigation of ozone treated and untreated aerosols revealed the formation of oxygen containing groups on the surface, indicating that the increased hydrophilic behavior of the investigated carbon aerosols is not due to soluble reaction products, but due to a significant change of surface properties. CONCLUSION The condensation properties of ultrafine carbon particles in the Aitken range were investigated by means of a variable supersaturation condensation nucleus counter. Whereas untreated carbon aerosols exhibited an extraordinary strong hydrophobic surface (contact angle of 50~ the ozone treatment resulted in a significant more hydrophilic behavior (contact angle of 26 ~ of the particles. As the same contact angle was found for all particle sizes and ozone concentrations, the change of the surface properties dominates the condensation properties of the particles, while the agglomerated particle structure is of minor importance for these particles. Critical supersaturations could not be lowered beneath 10 % by ozone treatment, therefore, it seems highly unlikely that these aerosols can act as condensation nuclei. The results indicate that the atmospheric residence time of soot is mainly determined by the surface chemistry of the particles. REFERENCES Colbeck, I., Appleby, L., Hardman, E. J. and Harrison, R. M. (1990) The Optical Properties and Morphology of Cloud-processed Carbonaceous Smoke. J. Aerosol Sci. 21,527-538. Crouzet, Y. and Marlow, W. H. (1995) Calculations of the Equilibrium Vapor Pressure of Water over Adhering 50-200-nm Spheres. Aerosol Sci. Technol. 22, 43-59. Fletcher, N. H. (1958) Size Effect in Heterogeneous Nucleation. J. Chem. Phys. 29, 572-576. Hagen, D. E., Trueblood, M. B. and White, D. R. (1989) Hydration Properties of Combustion Aerosols. Aerosol Sci. Technol. 10, 63-69. Hallett, J., Hudson, J. G. and Rogers, C. F. (1989) Characterization of Combustion Aerosols for Haze and Cloud Formation. Aerosol Sci. Technol. 10, 70-83. Helsper, C., MSlter, W., LSffier, F., Wadenpohl, C., Kaufmann, S. and Wenninger, G. (1993) Investigations of a New Aerosol Generator for the Production of Carbon Aggregate Particles. Atmos. Env. 27A, 1271-1275. Pruppacher, H. R. and Klett, J. D. (1978) Microphysics of Clouds and Precipitation. D.Reidel, Dordrecht. Stephens, S., Rossi, M. J. and Golden, D. M. (1986) The Heterogeneous Reaction of Ozone on Carbonaceous Surfaces. Int. J. Chem. Kin. 18, 1133-1149.
REDISTRIBUTION MECHANISMS FOR VOLATILE AND NON-VOLATILE COMPONENTS OF ATMOSPHERIC AEROSOL PARTICLES
H. Bunz, M. Koyro, O. Miihler, H. Saathoff Forschungszentrum Karlsruhe GmbH Institut ftir Meteorologie und Klimaforschung Postfach 3640, D-76021 Karlsruhe, Germany
Abstract - Atmospheric aerosols exist in the form of external nents like e.g. NH4NO 3 are usually found internally mixed experimental and theoretical study the responsible processes on temperature, relative humidity and aerosol parameters are
or internal mixtures. Volatile compoin the coarse mode. In a combined are identified and their dependencies determined.
Keywords - condensation/evaporation, coagulation, mixed aerosols, atmospheric aerosols
INTRODUCTION Two major components of the atmospheric aerosol are NHnNO 3 and (NH4)2SO 4. Usually NHaNO 3 as a quite volatile component is enriched in the larger particles (Bassett and Seinfeld, 1984, Weisweiler and Schwarz, 1990). Therefore, the interesting question arises if a particle system with a just opposite distribution of the two components is stable or if redistribution will take place. Two mechanisms for the redistribution compete with each other, coagulation and evaporation with subsequent condensation. Of course, the last process is important only for volatile materials as e.g. NH4NO 3 or some organic compounds. To quantify the amount of material redistributed by each of the two processes laboratory experiments were carried out using a system with one non-volatile ((NH4)2SO4) and one volatile (NH4NO3) component (Saathoff et al., 1995) and for comparison a second system with two nonvolatile components (NaNO 3 and (NH4)2SO4). At the same time calculations with the computer model NACHE (Bunz et al., 1995) were performed to analyze the process in more detail and to recalculate the experiments. Since the particle size distributions generated experimentally overlap to some extent, some additional modelling investigations were carried out on the basis of two very narrow and not overlapping size distributions. These calculations can help to understand to which extent under which conditions which process is responsible for the redistribution. EXPERIMENTS The experiments are performed in the 4.5 m 3 aerosol chamber ASA (Fig. 1) (Haury et al., 1978). Typical experimental duration is 12 to 24 h depending on the type of aerosol and other parameters especially the relative humidity. The mixing of the particles within the chamber is achieved by a small thermal gradient causing some natural convection. The chamber is equipped with sensors to control and to measure temperature, pressure and relative humidity. The particles were generated by dispersing a solution of 1.0 wt.% NH4NO 3 or NaNO 3 respectively in water with synthetic air through a nozzle (Schlick, Type 970) and subsequent drying in a diffusion dryer. The (NH4)zSO4-particles were produced using an ultrasonic nebulizer (Sinaptec, Type EGA2400) containing a solution of 10 wt.% (NH4)2SO 4 in water. Again the particles are dried by a diffusion dryer. In both cases the particles are neutralized by a Kr 85 source and their coarse part is removed by
469
470
Bunz et al.
a cyclone. The distribution parameters achieved by these methods are rg = 0.15 [am for the nitrate particles and rg = 0.7 [am for the sulfate particles with a standard deviation cy = 1.6 in both cases. The evolution of the particle number distribution is measured by a mobility analyzer (TSI, Type 3071) for the particles between 0.02 [am and 0.85 [am in diameter and by an opticle particle counter (Pallas, PCS 2000) in the range between 0.3 grn and 16 grn. To obtain the composition of the particles as a function of their size, probes are taken at certain times using a nine stage low pressure impactor (Berner et al., 1979) and analyzing the samples by ion chromatography (Dionex, Type 2010i). These impactor measurements are the basis for the subsequent analysis of the experiments by the model calculations. As additional integral values the total number concentration is measured with a condensation nucleus counter (TSI, Type 3020) and the total mass concentration with filter probes. AEROSOL PHYSICO-CHEMICAL MODEL The computer model NACHE (Bunz and Dlugi, 1991; Bunz et al., 1993) is able to handle an arbitrary number of chemical components assuming that all particles in a specific size class are equally composed and that the composition varies only as a function of the size. If a component is specified as an electrolyte, the correct dissociation products are provided and the corresponding libraries are loaded which contain activities, densities and other material properties as a function of the ionic strength and temperature. On the basis of the activities of the different electrolytes and the water the amount of solved material and of the water in the droplets can be calculated as well as evaporation and condensation if the corresponding activity in the gas phase is known. The activity in the gas phase can be determined if data for the vapour pressure of the material under consideration are available either for the pure substance or for some standarized state (e.g. m=l). The semi-analytical routine in NACHE originally developed to compute condensation of H 2 S O 4 o n the particles was modified so that condensation/evaporation of any component can now be calculated. At the moment vapour pressures for H z S O 4, H 2 0 and a quasi-vapour pressure of NH4NO 3 based on the dissociation constant (Stelson and Seinfeld, 1982, Mozurkewich, 1993) are implemented in the code. As long as NH 3 and HNO 3 are equimolar in the gas phase this "vapour pressure" of NH4NO 3 is equal to the square root of the dissociation constant. In our experiments it is assumed that this condition is fulfilled and that NH 3 and HNO 3 can be treated as gaseous NH4NO 3. Whereas the activity of H2SO 4 is usually very small so that only condensation occurs under nearly all circumstances, the activity of NH4NO3 in the particles can be lower or higher than the activity in the gas phase depending on the composition and the physical state (liquid or solid) of the particles. Smaller particles evaporate faster or grow slower due to the Kelvin-effect. To quantify the partitioning between the two processes responsible for the redistribution, coagulation and condensation/evaporation, a tool is implemented in the code to "measure" the flow of material under consideration from the shrinking part of the particle size distribution to the growing part. RESULTS Two series of calculations are presented, both are closely related to the experiments in the ASAchamber. They differ in the particle size distribution used to start the calculations. In the first series the initial size distribution is chosen in a way so that the effects of the redistribution can be separated more easily. The second series is started with the experimental initial size distribution. The parametric calculations are performed with an initial very well separated bi-modal distribution, the 1st mode consisting of NH4NO 3 particles (rg = 0.1 [am, Cyg = 1.1, M o = 1 mg) and the 2rid mode of (NH4)zSO 4 particles (rg = 0.5 [am, (Yg -" 1 . 1 , M 0 -~- 10 mg). The relative humidity (r.h.) is varied between 40% and 85%. and the temperature between 5C and 45C. The effective accommodation coefficient is chosen to be 0.3 for the dry case (below the deliquescence point, r.h.> 1 KT corresponds to the case of"highly volatile" particles, and for A we have 3 1 A ---2 I + - - - - c~ - +sT 1 tr 7" g r a d T k r cTT
2 Nk r
THERMOPHORETIC VELOCITY The velocity field around the sphere obeys the Navier-Stokes equations. From boundary conditions for tangential and normal components of gas velocity at the particle surface we obtain v
o%' A sin 0
The spherically-symmetric component of v governs the mass flux onto the particle s T In view of one of the former equation we obtain the familiar relation I = -4 ~---x b'T L Non-symmetrical component of v contributes to the thermophoretic force. The steady particle motion with the velocity vvp (thermophoretic velocity) corresponds to the condition that the pressure force of the gas onto the particle equal to zero. As a result we have vrp - - -3 A krs --f + DkDs cTF + PsL "
+
+ ~,L--f- gradT
For particles with high thermal conductivity the effect of volatility is governed by the last term of the equation. It does not depend on thermal conductivities of gas as well as of particle and can be essentially described as the formation around the particle of a gas mixture affecting the thermophoresis by providing an uncompensated momentum due to thermal diffusion. We computed the thermophoretic velocity for mercury and NaC1 aerosols in the air in presence of water vapor. The latter can condense and evaporate from particle surfaces. It can be seen that water volatility at room temperature results in increasing the thermophoretic velocity for the aerosols considered only 23 times. Even if the temperature is of the order of 50 C the velocity can increase only 10 times compared to Epstein's value. Thus the experimentally observed values of thermophoretic velocity cannot be accounted for solely by the effect of volatility.
492
Bakanov
CONCLUSIONS 1. Due to the effect of volatility 0 the temperature distribution inside and outside the particle changes; iO the surrounding space fills with the vapor of the volatile substance leading to thermal diffusion in the mixture and an uncompensated transfer of momentum from gas molecules to the particle. 2. At room temperature the change in the temperature field is more essential for particles wi}h low thermal conductivity. 3. On the contrary, at the same conditions the thermal diffusion effect is more essential for particles with high thermal conductivity. 4. The fact that the experimentally obtained values of thermophoretic velocity significantly exceed the computed ones (by Epstein's formula) cannot be accounted for solely by the effect of volatility. DESIGNATIONS: v - velocity, T- temperature, p - pressure, p - density, D - coefficient of mutual diffusion, tc - thermal conductivity, v - cinematic viscosity coefficient, R - aerosol particle radius, krs - creep coefficient, k~)s - diffusion sleep coefficient, k r - thermal diffusion ratio, L - latent heat, C - concentration of the volatile component. INDEX NOTATIONS: c - center of gas molecule mass, i - condensed phase, r - radial component, 0 - tangential component, 1 o r 2 - volatile or non-volatile component, s- saturated vapor, o- center of sphere
POLYCHLORINATED
BIPItENYLS IN TIlE AMBIENT AIR OF SOUTHERN TAIWAN
Shui-Jen Chen*, Lien-Te Hsieh and Ping-Shium Hwang Department of Environmental Protection Technology National Ping Tung Polytechnic Institute Nei Pu 91207, Ping Tung, Taiwan
Keywords - polychlorinated biphenyls, concentration, phase distribution, particle size distribution, urban, rural, industrial, ambient air
Abstract - PCB (Polychlorinated Biphenyl) samples in the ambient air of rural, urban and industrial sites have been collected by two PS-1 samplers and two MOUDIs (Micro-orifice Uniform Deposit Impactors) from July, 1993 to December, 1994 in southern Taiwan. Total-PCB concentrations averaged 2.50, 4.51 and 5.91 ng/m 3 for rural, urban and industrial sites, respectively. Mean gas phase distribution of total-PCBs was 43.6%, 60.9%, and 63.1% for the rural, urban, and industrial sites, respectively. The patterns of PCB homologues were quite similar with the distribution of those found in the mixture of Aroclor 1242 and 1260. Particle-bound total PCBs averaged 10.3, 13.9 and 9.24 ~g/g for rural, urban and industrial sites, respectively. For both urban and industrial sites, the particle size distribution (dC/dlogDp vs. Dp) of both total-PCBs and total-particle mass was found to have bimodal size distribution. The industrial aerosols are dominant in the fine particle mode (Dp 2.5 ~m). INTRODUCTION Polychlorinated Biphenyls (PCBs) are primarily used in transformers and capacitors as dielectric fluid because that they have a heavy oil-like consistency, a high boiling point, a high degree of chemical stability, low electrical conductivity, low flammability, and a specific gravity between 1.20 and 1.44. PCBs are also used in a variety of other applications such as: heat transfer and hydraulic fluids; dye carriers in carbonless copy paper; plasticizer in paints, adhesives, and caulking compounds; and fillers in investment casting wax (Toxic Information Series 1993). PCBs exhibit lipophilic and hydrophobic properties in water and therefore accumulate in lipid layers of biota. The ability of PCBs to volatilize from landfills, and resist degradation at low incinerating temperatures makes atmospheric transport the primary mode of global distribution. Due to the fact that PCBs-containing products are used widely in urban areas, the PCB concentrations in the ambient air of urban areas is one order of magnitude higher than those in rural or remote areas (Eisenreich et al. 1981;Doskey and Andren 1981;Murphy et al. 1985; Swackhamer and Swackhamer 1986; Swackhamer and McVeety 1988;Manchester-Neesving and Andren 1989). PCBs have been observed in soil, water (Brown 1985) and sediments (Swackamer 1986), as well as fish (Muir 1988). In addition, several studies have measured the concentration of PCBs in rain (Murphy 1977; Duinker 1989), snow (Murphy 1983) and ambient air (Doskey 1981; Bidleman 1987; Manchester-Neesvig 1989; Baker and Eisenreich 1990).
* Corresponding author
493
494
Chen et al.
In Taiwan, many of the PCBs previously used are still in service (Wu 1989). In 1989, the disposal of PCBs was strictly regulated by the Taiwan Environmental Protection Agency (EPA 1993). Prior to 1989, however, many PCB-containing wastes were disposed in municipal landfills or at uncontrolled waste disposal sites (Wu 1989). Due the fact that PCBs possess sufficiently high vapor pressures (104-- 104 Kpa) to be emitted directly into the air from the surrounding hazardous waste disposal sites or through volatilization from contaminated surfaces (Weaver 1984), they also may be released from controlled landfills through vents along with more volatile gases. Because of energy consumption and emergency use, self-power plants are essential to industrial plants. Therefore, the industrial plant is also a possible source of PCB emission. Atmospheric transport is an important pathway for the transfer of PCBs from a ground-base to a higher-elevation site. In the past, many studies focused primarily on PCBs in the gas, fine-particulate and coarse-particulate phase. However, there is little information available on the comparison of different sampling heights for PCBs in the field. The main objectives of this study are to investigate the characteristics of PCBs in the ambient air of urban, industrial, and rural areas associated with concentration, phase distribution, particle-bound PCB composition, size distribution, and height effect. In order to obtain the above objectives, twenty-six PCB samples in the ambient air of urban were taken from July 1993 to Oct 1994. In addition, 13 and 20 PCB samples in the ambient air of industrial and rural sites, respectively, were determined and compared with the urban samples. EXPERIMENTAL METHODS Semi-Volatile Sampler Ambient air samples for both particle and the gas phases of PCBs were collected using several standard semi-volatile sampling trains (General Metal Works PS-1). The PS-1 sampler with a glass fiber filter (cleaned with distilled-deionized water and heated at 450~ ) was used to collect total suspended particles (TSP) and the particle phase PCBs. A glass cartridge containing a 5 cm polyurethane form (PUF) plug followed by a 3 cm XAD-2 resin, and finally a 2 cm PUF plug was used to collect the gas phase PCBs. The PUF/XAD-2 cartridge was cleaned by sequential extractions. The glass fiber filters were weighed before and after sampling to determine the amount of particles collected. PUF/XAD-2 cartridges were stored and transported in clean screw-capped jars with Teflon cap liners. Glass fiber filters were transported to and from the field on a prebaked glass plate and wrapped with aluminum foil. Microorifice Uniform Deposit Impactor (MOUDI) A microorifice uniform deposit impactor (MOUDI) was used to measure size distribution of particle and PCBs in the ambient air of both industrial and urban sites. The MOUDI (MSP Corporation Model No. 100) consists of two basic assemblies. One is the cascade impactor and the other is the rotator (MSP Corporation 1989). The impactor consists of eight or ten stages, plus an inlet and an afterfilter located in the base. At each stage the collected particles are deposited uniformly over the entire impaction plate by the corret radial placement of nozzles. The impactor was operated at 30 L/min and the nominal cut-size of the impactor stages are as follows : 0.056, 0.1, 0.18, 0.32, 0.56, 1.0, 1.8, 3.2, 5.6, 10 and 18 gm (Marple et al. 1991). The rotator can rotate alternate sets of upper impactor/nozzle assemblies to each other and the nozzle plates are rotated relative to the impaction plates to achieve a near uniform particle deposit. Two types of filter were used as substrates in the impactor. One was aluminum foils with 37 mm diameters for each impaction stages from MSP Corporation. The other was the Zefluor filter in 37 mm diameters with mesh 2/xm( Gelman Sciences Co.) for the after-filter base. Both aluminum foils and Zefluor filters were prewashed by a solvent solution (a mixture of n-hexane and dichloromethane, V:V = 300 mL/L: 300 mL/L) for 24 hours. In order to reduce particle bounce during sampling, the aluminum foil filters was coated with a light grease to form a sticky surface. After the grease has been applied, it is necessary to bake the aluminum foil filters in an oven at 60~ for 90 minutes. These
Polychlorinated biphenyls in the ambient air of Southern Taiwan
495
filters were weighed before and after sampling to determine the amount of particles collected. The MOUDI has removable impaction plates that can be quickly interchanged between different runs. Substrates can be installed on the impaction plates in the laboratory. The plates were transported to the field, removed from the impactor after a sampling and transported back to the laboratory for the substrate removal and analysis. Covers are used to prevent the substrates from contamination. PCB Analysis After final weighing, the PCB sample was placed in a solvent solution (a mixture of n-hexane, acetone and dichloromethane, V:V:V = 300 mL 9150 mL 9150 mL, respectively), and extracted in a Soxhlet extractor for 24 h. The extract was then concentrated, cleaned-up and reconcentrated to exactly 1.0 or 0.5 mL using a procedure similar to those described by Manchester-Neesving (1989) and Lee (1991). A gas chromatograph (GC)(Hewlett-Packard 5890 Series II Plus) with a 63Ni electron capture detector (ECD) and a computer workstation was primarily used for the PCB analysis. This GC was equipped with a Hewlett-Packard capillary column (HP Ultra 2 - 50 m X 0.32 mm X 0.17/xm), an HP-7673A automatic sampler, an Electronic Pressure Control(EPC). The Helium carrier gas flow is 0.9 mL/min at 285~ The GC conditions for PCBs analysis were as follows: splitless injection (270~ the volume of sample injected 2 /xL, detector-ECD 300~ oven 50~ to 100~ at 3~176 to 240~ at 0.6~ 9240~ to 290~ at 4~ hlod at 290~ for 55 min. A 95% argon and 5% methane mixture was used for makeup gas. The GC for PCB analysis was calibrated with a diluted equal mass mixture of Aroclor 1242, 1248,1254, and 1260. Concentrations of the individual congeners were calculated using data from Manchester-Neesving and Andren (1989). A mixture of 10 congeners was used to generate a correction curve to Mullin et al.'s relative retention time (RRT) data (Mullin et al. 1984) for individual congener described in Manchester-Neesving and Andren (1989). Specific PCB congeners identified, grouped by homologue and used in this study are similar to those described by Lee(1991). Standards were injected daily to confirm retention times and response factors. Analysis of serial dilutions of PCB standards found that the limit of detection was between 0.8 and 71 pg for individual congeners in previous study. The limit of quantification (LOQ) averaged approximately 0.4 pg/m 3 for semi-volatile sampler. Ten consecutive injections of a PCB standard yielded an average calculated error (standard deviation/average area) of 0.133 with a range of 0.015 to 0.183. Recovery of internal standards (octachloronaphthalene (OCN) and tetrachloronaphthalene (TCN)) injected into the extraction solvent averaged 0.912+0.041. Recovery of OCN applied directly to the filter before sampling averaged 0.83+0.05 (Lee 1991). PCBs recovery efficiencies were determined by processing a solution containing known PCBs concentrations through the same experimental procedure used for the samples. The recovery efficiency for individual PCB congeners injected into the extraction solvent averaged 0.86 (Lee 1991). The blank tests for PCBs were accomplished by using the same procedure as the recoveryefficiency tests without adding the known standard solution before extraction. Analyses of field blanks, including filters and PUF cartridges, found no significant contamination (GC integrated area < detection limit). Analysis of duplicate experiments yielded differences in total-PCB concentration averaging 0.083 for ambient air samples. Sampling Program The PCB samples in the ambient air at the industrial, urban, and rural sites were collected from July 1993 to December 1994 in southern Taiwan. Industrial site samples were taken on both the roof of a three-story building (12 m height) and the ground nearby the building (1.5 m height) located at the center of an industrial park in southern Taiwan. Urban samples were taken on both the roof of a four-story building (12 m height) and the ground near the building (1.5 m height) located in a mixed institutional, commerical, and residential area in the center of Ping Tung City. This building is on the Lu-Shing Junior High School Campus. Rural samples were taken on both the roof of a three-story building (12 m height) and the ground near the building (1.5 m height) located on the National PingTung Polytechnic Institute campus. National Ping-Tung Polytechnic Institute, which is located 10 km south of Ping-Tung City and 2 km west of a small town - Nei-Pu, has a large farm, greensward and several surrounding hills which provide a typical sampling-site for the rural area.
496
Chen et al.
RESULTS AND DISCUSSION Mean total-PCBs concentration in the ambient air was 5.91, 4.51, and 2.50 ng/m 3 for the industrial, urban, and rural sites, respectively. This result indicates that PCB concentrations in the ambient air might correlate with human and/or industrial activities. Mean concentration of each PCBhomologue was quite similar to found in the mixture of Aroclor 1242 and 1260. This result reveals that the ambient air of urban and industrial sites in southern Taiwan had been contaminated by PCB wastes. Mean value of total-PCBs composition for the industrial, rural, and urban sites indicates that the more volatile PCBs were dominant in the gas phase, while the higher molecular weight PCBs were primarily associated with particulates. An evaluation of the G/R ratio of total-PCB concentration in the three sampling sites shows that those differences may be influenced by the sources of pollutants and the local meteorological conditions. In general, the total-PCB concentration in both urban and rural sites decreased with increasing in the height of sampling site, but it increased with increasing the height of sampling elevation for the industrial site. For both urban and industrial sites, the particle size distribution (dC/dlogDp vs Dp) of totalPCBs was found to have bimodal size distribution. This result indicates the typical dispersion and depletion of total-PCBs from the ambient air of industrial site to the urban site. and reveals the phenomena of condensation process for the young aerosols. In the urban site, a greater fraction of PCB mass existed in the coarse aerosol, while, in the industrial site, most of PCB mass existed in the fine aerosol due to the local characteristics with many asphalt roads and fewer traveling vehicles in this area. REFERENCES
Baker, J.E. and Eisenreich, S.J. (1990) Environ. Sci. Technol. 24, 342-352. Bidleman, T.F., Wideqvist,U., Jansson, B., Soderlund, R. (1987)Atmos. Environ. 21, 641-654. Brown, M.P., Werner, M.B., Sloan, R.J., Simpson, K.W. (1985)Environ. Sci. Technol. 19,656-666. Doskey, P.V. and Andren, A.W. (1981) Great Lakes Research. 7, 15-20. Duinker, J. C. and Bouchertall, F. (1989) Environ. Sci. Technol. 23, 57-62. Eisenreich, S.J., Looney, B.B. and Thornton, L.D. (1981) Environ. Sci. Technol. 15, 30-38. EPA, (1993) Environmental regulations.Taiwan, Environmental Protection Information Inc. EPA, (1993) Toxic chemical substances PCBs management handbook. Taiwan, Environmental Protection Agency. Lee, W.J. (1991) The determination of dry deposition velocities for ambient gases and particles. Ph.D. Thesis. Chicago, IL, 60616: lllinois Institute of Technology. Manchester-Neesving, J.B. and Andren, A.W. (1989) Environ. Sci. Technol. 23, 1138-1148. Marple, V.A., Rubow, K.L. and Behm, S.M. (1991) Aerosol Science and Technol. 14, 434-446. Muir, C.G., Norstrom, R.J. and Simon, M. (1988) Environ. Sci. Technol. 22, 1071-1079. Mullin, M.D., Pochini, C.M., McCrindle, S., Rornkes, M., Safe, S.H., Safe, L.M. (1984) Environ. Sci. Technol. 18, 468-476. Murphy, T. J., Formanski, L.J., Brownawell, B., Meyer, J.A. (1985) Environ. Sci. Technol. 19, 942-946. Murphy, T.J. and Rzeszutko, C.P. (1977) Journal of Great Lakes Research. 3, 305-312. Murphy, T.J. and Schinsky, A.W. (1983) Journal of Great Lakes Research. 9, 92-96. Swackhamer, D.L. and Armstrong, D.E. (1986) Environ. Sci. Technol. 20, 879-883. Swackhamer, D.L., McVeety, B.D. and Hites, R.A. (1988) Environ. Sci. Technol. 22, 664-672. Toxic Information Series. (1993) Polychlorinated Biphenyls. Washington, DC 20460: U.S EPA Toxics Information Series United States Environmental Protection Agency, Office of Toxics Substances, TSCA Assistance Office (TS-799). Weaver, G. (1984)Atmos. Environ. 18: 22A-27A. Wu, S.C. (1989) SINO-US BI-National Conference on Environmental Protection and Social Development, Taipei, Taiwan.
SIZE DISTRIBUTION OF IONS IN A T M O S P H E R I C AEROSOLS Z. KRIVACSY AND A. MOLN,M~ Air Chemistry Group of Hungarian Academy of Sciences at University of Veszprrm, H-8201 P.O.Box 158, Veszprem, Hungary Abstract - The aim of this paper is to present the size distribution of inorganic cations (ammonium, sodium, potassium, calcium, magnesium) and anions (sulfate, nitrate, chloride) as well as of some carboxylic acids. The results obtained show that, beside of ammonium, sulfate and nitrate, organic ions are in the fine particle size range. Potassium and chloride are rather uniformly distributed between fine and coarse particles. Further, carbonate, sodium, calcium and magnesium ions are mostly found in the coarse size range. Re-calculating the concentrations in equivalents a cation excess is found with a minimum between 0.5 and 1.0/zm. Keywords - size distribution, particle origin
INTRODUCTION Water soluble aerosol particles play an important role in the control of several atmospheric processes including light attenuation, cloud formation, etc. The effect of particles depend on their composition as well as the size distribution of different compounds in the atmospheric aerosol particles. For this reason many atmospheric observations were carried out to determine these parameters. Thus, in Hungary the size distribution of different inorganic ions were studied many years ago (Mrsz~iros, 1968; 1970). Owing to environmental changes since that time these measurements are recently repeated by using much more reliable methodology. Into these new investigations the identification of some organic anions are also included. EXPERIMENTAL The aerosol samples were collected on aluminum foils by using of 8-stage Bemer-type impactor. With a sampling rate of 31.2 l/min the aerodynamic cut-off diameters of the stages are 0.0625, 0.125, 0.250, 0.50, 1.0, 2.0, 4.0, 8.0 and 16 ttm . Sampling was made on the roof of a building of the University of Veszprrm, about 20 m above the street level. Veszprrm has a population of 60000, in the town the strength of industrial sources can be neglected. The duration of the sampling was about 60 hours which corresponds to an air volume of 120 m 3. Between September 1995 and February 1996, 12 samples were taken. The samples on the aluminum foils were extracted in ultrasonic bath with 10 ml of Milli-Qwater for 30 minutes. The water used was boiled for 30 min before the extraction to improve the reliability of pH-measurements and the determination of carbonate concentration of atmospheric aerosol. The ionic composition was analysed by free solution capillary electrophoresis. For this purpose Waters Quanta 4000 capillary electrophoresis system was used with fused silica capillary of 75 ttm inner diameter and 60 cm length. The pH of the aerosol samples was determined and the result was corrected to the blank water used for extraction.
497
Krivdcsy and Moln6r
498
RESULTS The average total and fine concentration of different organic and inorganic ions is presented in Table 1. It can be seen that anions are dominated by sulfate and nitrate and in lesser way chloride, as expected. The concentration of organic ions is much lower than that of inorganic species. In this group oxalate is the most abundant ion. In the cations ammonium has the greatest concentration. The concentration of potassium is surprisingly high as compared to the other cations considered generally of soil origin. The table also shows that, except chloride and carbonate, all anions measured are dominantly in the fine size range, while some cations (Ca, Na, Mg) are mostly detected in the coarse size range. It is rather surprising that the majority of potassium and an important part of sodium is found in the range of fine particles. Heintzenberg (1989) compiled chemical data for fine particles. According to his work the total mass concentration of fine particles under non-urban continental conditions is 15 #g m-3,, while sulfate, ammonium and nitrate ions give 37, 11 and 4% of the total mass. This means that their concentrations 5.55, 1.65 and 0.6/~g m -3, respectively. This means that our ammonium and mostly nitrate concentrations are higher than overall averages. Table 1: Average total and fine (d < 1 ~m; in parenthesis) concentration of different anions and cations in/~g m -3. Chloride Sulfate 1.5 7.1 (0.75) (4.3)
Nitrate 4.5 (2.9)
Carbonate Oxalate 0.33 0.13 (0.06) (0.08)
Ammonium 4.2 (2.6)
Potassium 1.8 (1.0)
Calcium 0.6 (0.14)
Format 0.03 (0.02)
Succinct 0.03 (0.02) Sodium 0.48 (0.21)
Malonate 0.03 (0.02)
Acetate 0.01 (0.01)
Magnesium 0.07 (0.02)
Fig. 1 shows the size distribution of the mass concentration of the more important anions and cations. One can see that both sulfate, nitrate and ammonium have very similar size distributions with a maximum between 0.5 and 1.0/~m. The figure indicates that oxalate ions are also distributed in the same way. However, potassium and chloride have very different size distribution. Their mass concentration is nearly equally distributed in the fine size range (d--~ 0.2/tin), algae, spores of lichen, mosses, ferns and fungi (r >,-~ 0.5tim), pollen (r >,~ 5ttm) (Macher, 1993), plant debris like leaf litter, parts of insects, human and animal epithelial cells (supposed r > l#m). They are an ubiquitous component of the atmospheric aerosol and come to about 24% of the concentration of the total atmospheric particles (Matthias-Maser and Jaenicke, 1995). Besides their effects on air hygiene they play an important role in cloud physics. Schnell and Vali, 1973, suggested that a portion of atmospheric freezing nuclei was of biogenic origin. The sources of these nuclei include decaying vegetation, marine plankton (Schnell and Vali, 1976) and bacteria (Maki and Willoughby, 1978, Levin et al. 1980). The peculiarity of the bacteria is that they have a freezing capability even at temperatures about - 4 ~ while most mineral particles need temperatures below -10 ~ C. Ice nucleation activity of the free living fungus Fusarium was mentioned by Pouleur et al, 1992 . Even pollen can contribute to condensation processes. Durham, 1943, worked with 12 species of pollen to determine their response to air of various humidity levels. He found that all were hygroscopic in the sense that they acquired water from tile vapor below 100% relative humidity. The species studied include six weeds, two grasses and four deciduous trees. Dingle, 1966, mentioned the hygroscopicity of the protoplasm of ragweed (ambrosia artimcsiifo-
526
Size distribution of primary biological aerosol particles in rain-water
lia) at humidities higher than 51%. In addition, primary biological aerosol particles are insoluble so they are expected to be present in rain-water samples.
2
Methods
In our study the rain-water samples were taken in Mainz, all urban/rurM influenced region during a period from .January to March 1995. Tile salnpling site was the botanical gardeI~ of tile university of Mainz. For collecting rain water an automatic "wet-only" rain-watei sampler was used, whose cover was controlled by a humidity sensor in order to ~void contaminations from dry deposition. After filtration, the rain-water was investigated in a light microscope (giant particles, r > 2#rn), and tile large particles (0.2 > r > 2tim) were examined in a scanning electron microscope (SEM) equipped with all energy dispersive Xray st)ectrometer (EDX). For classification the giant particles were stained with a proteitl dye (Matthias-Maser and .laenicke, 1994). This was done by adding 0.05 ml of the stainiilg solution per 20 ml of rain water. After about 10 days of storage at 1~ (in darkness t~ l)revent tile samples from mould) most of the biological particles show a bl,lish cc~l,~,l, (Gruber, 1995). In SEM the biologicM particles show a striking morphology (spheres. rc~,Is. characteristic forms) together with a special elemental composition (P, S, t(, ('a, sometimes Si, CI on a high background spectrum) and some of them change their form during E[)X (shrinking) (Matthias-Maser and .Jaenicke, 199,5).
3
Results
After measurenlent the particles have been counted, classified and the size distrib,ltions c~f ttle insoluble aerosol particles, total and biological, were determined. Within tile different size distributions during the sampling period the starting of tile pollen season in Feb./ March is evidently seen. The percentages of PBAP in tile corresponding size ('lasses came up to 90% of the total aerosol. Considering all measurements the mean size distributions of tile total and tile biological aerosol particles in rain-water were calculated. These are shown in fig. 1 together with the percentages of biological particles in each size class. With increasing particles radius tile particle concentration decreases more or less. The fraction of PBAP varies between the different particle sizes. Although it is known that eg. bacteria are good freezing nuclei (see introduction) only very few large particles were biological. Most water insoluble particles of this size consisted of minerals. Tile concentration of PBAP shows a second maximum in the size classes of tile mean radii 11.41tin and 14.7#m. This is caused by tlle starting pollen seasons (hasel and alder ) in Febuary. Regarding tile entirety of insoluble particles, tile fraction of biological particles were 9% of the total particles with a mean number concentration of 3.5. 103rn1-1 (Gruber, 199,5). This seems rather low in contrast to the values found in tlle atmospheric aerosol (24%), but it is not easy to compare these two concentrations. Once suspended in water, many partially soluble particles decay in several smaller particles. The result is an increase of the number concentration of the non biological particles in the smaller size classes. This leads to a smaller percentage of the biologi~'al particles' n,lmber concentration to the total particles. In view of tile fact that the giant
527
528
M a t t h i a s - M a s e r et al.
biological particles are more a b u n d a n t ill tile distribution of particles in rain-water it is useflll to regard their volume cotlcentration respectively their mass. Calculating the mean values of the unsoluble mass in rain-water the total mass comes to t.5 mg/l (with mean particle density ~ = 1.59/cm 3) and a biological mass amounts 0.,5 rag/1 (mean particle density -fiP~AF' = 19/cm3) 9 This corresponds to 33% of the total mass. The biological (;ontent however is strongly dependent on the seasonal variation because of tile phenology. Additional m e a s u r e m e n t s in other seasons could lead to other results.
10 5
-C total aerosol
N N
- - - - - - - biological particles
N
%__.
......
...x_
10 4
percentage
._,,._
10 3
"-
10 2
E c-
'
o -
k._
url o
10
-- 100
J-s0
1
Z "72J
100
t
." 9 ,' 9
.'
,
40
', ,
R
9
e
,
9 9
t
o
20
,
o o
10 .2
regen
0,1
~ E
0
10 -1
60
.
9
!
!
i
i
1
i
i
ii
l
10
i
i
|
i
i
i l l
0
100
rin gm Fig.l" Mean size distributions of the insoluble aerosol particles and PBAP in rain water.
Size distribution of primary biological aerosol particles in rain-water
References Dingle A.N. (1966), Pollen as Condensation Nuclei, .l.R.cch,. Atmo.sph., 2, p.231 - 2 3 7
(~96(~). Durham ().(1. (1943), The volumetric incidence of atInospheri(" allergenes, I. Specific gravity of pollen grams. J. ,411(:r'g9, 14, 6, p . 4 5 5 - 461. Gruber S.. (1995), Biologischer Anteil der unl/Sslicheil Bestan(Iteile von Partikeln im Niederschlag, [)iplomarbeit am Inst. f/ir Physik der Atm~st)hs der Universiti~t Mainz. IGAP (1994), A plan for international (i;lobal Aerosol Program, P. Hobbs (ed), Geneva, 1994. Levin Z., Sandle,mann N., .Xloshe A., Bertold T., Yankofsl 0.2#nz. J. Aerosol ,5'cie:tzcc Vol 25, No. S , p . 1 6 0 5 - 161:3. Pouleur S., Richard C., Martin J.-G., Anto,m H. (199"2), Ice N,lcleation Activity in Fusa,'ium acuminaturn and Fusarium avcnaccum, Applied a T~d Environmental Microbiolog9 Vol. 58, No. 9, p.2960 - 2964. Schnell R. C., Vali G. (1973), World-wide Source of Leaf-derived Freezing Nuclei,Nature London , 246, p.212 - 21:3. Schnell R. C., Vali (;. (1976), Biogenic Ice Nuclei: Part I, Terrestrial and Marine Sources, J. Atmosph. Sciences, :3:3, p . 1 5 5 4 - 1564.
529
Comparisons Between Hydrated and Dehydrated Aerosol Particle Size Distributions P.S.K. Liu, W.R. Leaitch and C.M. Banic Atmospheric Environment Service, 4905 Dufferin Steet, Downsview, Ontario, M3H 5T4, Canada
Abstract - Simultaneous observations of the aerosol size distribution, dried and in-situ, and aerosol ion chemistry made off the coast of southern Nova Scotia, Canada during 1993 are used in a preliminary examination of the growth of atmospheric particles by water absorption. When compared with a simple model for the uptake of water by sulfate particles, these observations suggest that sulfates alone could not account for the increase in measured particle size at different relative humidities. The addition of the measured organic and other inorganic ions notably improved the agreement between observations and the modeled results. Keywords - Aerosol, sulfate, organics and water absorption.
INTRODUCTION Changes in aerosol particle size due to water absorption results in changes in the particle cross section and scattering efficiency. This is a critical issue for the direct climatic effect of the atmospheric aerosol. Simultaneous observations of the atmospheric aerosol size distribution, dried and in-situ state were made during the 1993 North Atlantic Regional Experiment (NARE) off the coast of southern Nova Scotia, Canada from August 10 to September 8 1993. The observations were made from the National Research Council of Canada DHC-6 Twin Otter aircraft. These observations are used here to examine the size of the atmospheric aerosol as a function of relative humidity (RH). INSTRUMENTATION The real-time measurements pertinent to this study were made with two wing-mounted Particle Measuring Systems probes: a PCASP-100X and a FSSP-300. The PCASP uses laser light scattering to count and size particles in 15 size bins covering diameters from 0.13 to 3 l.tm diameter. The PCASP is equipped with an isokinetic diffuser cone to decelerate the aerosol sample from the aircraft speed before being counted. The aerosol is dried inside the diffuser section due to the applied heating, and also inside the detection area due to the dry sheath air and internal heat sources (Strapp et.al., 1993). The FSSP-300 also uses laser light scattering to count and size the in-situ aerosol from 0.3-20 l.tm. However, particle detection by the FSSP probes is not influenced by heat from the probe since particles pass through the laser detection area by the forward motion of the aircraft and are subjected to only minor flow perturbation by the probe geometry. The PCASP was size calibrated using near monodisperse particles of NaC1 and Latex generated by passing the atomized aerosol through a TSI electrostatic classifier. The FSSP-300 was calibrated with atomized Latex particles. Temperature and dew-point temperature were measured with a Rosemount 102 temperature probe and an E.G.&G. 137 dew-point hygrometer respectively. Aerosol particle chemistry was obtained from Teflon filter samples. The extracts (deionized water) were analyzed by ion chromatography for major inorganic and organic ions (Li et. al., 1996). The species measured are sulfate, nitrate, chloride, ammonium, sodium, propionate, acetate, formate, pyruvate, methane sulfonate, oxalate, glyoxalic acid, glycoaldehyde and formaldehyde.
530
Comparisons between hydrated and dehydrated aerosol particle size distributions 1 E+4 1 E+3
~ ~........
[
!
'__2 ~"
PCASP-100X 0
nO3 1 E+2 O
"lD
Z
-o
1E+1 1 E+O 1E-1
0.1
........
l
......
1.0 Diameter (grn)
531
OBSERVATIONS Data collected simultaneously with chemical filter sampling below 1.5 km are the focus of this study. The size distributions were averaged over periods for which the RH varied by no more than 5%. Since the RH usually varied by more than 5% during each filter sample, which may last for about 1 hour, more than one size distribution average was often obtained for each filter period. A maximum of 3 distributions averaged at different RH were analyzed per filter. The sample flow rate in the i0.0 PCASP is corrected for known variations due to altitude change.
Figure la Size distributions from PCASP-
Figure 1 are examples of particle size distributions measured by PCASP and FSSP for flights 24 and 25. 100X and FSSP-300 for flight 24b (RH 50%) Sample 24b, 25a and 25b were collected at 50%, 71% and 86% RH respectively. The FSSP-300 distributions are 1E+5 ........ ~ ......... shifted towards larger sizes compared to the PCASP distributions. These shifts, are assumed to be due to water 1E+4 addition. The difference in the particle size shift can also be seen between the three samples; larger particles are c3 1E+3 measured with the FSSP-300 for higher RH. ot~ The complete data set was analyzed to obtain the ~_. 1E+2 wet size of the particle for a corresponding dry diameter Z -o 1E+1 (do) of 0.4 gm. The wet diameter (d) is just the FSSP-300 d i a m e t e r c o r r e s p o n d i n g to the same d N / d l o g D
v a l u e as f o r
1E+0
. the PCASP at do. The choice of the dry diameter (do= 0.4 "' "',, gm), was governed by the desire to have a point well 1E-l.i ........ 1.01 . . . . ~.,,,10.0 within the detection range of FSSP-300 and one that Diameter(urn) represents the accumulation mode. Inherent in this analysis is the assumption that the particles are internally Figure l b Size distributions from PCASP mixed and the relative increase in size is assumed to be and FSSP-300 for flight 25a (RH 71%) due to water addition. There were 29 cases selected from 18 filters periods. Six cases for which no filters were collected have been added to increase the RH range. 1 E+4 ...... ! These are shown as open circles in the figures. The d/do vs RH plot obtained from the above procedures are shown 1 E+3 in Figure 2. Because the FSSP-300 was calibrated with o 1E+2 particles having a refractive index of 1.585, and the o particles being measured may have a smaller refractive Z -o 1E+1 -~ index because of water addition, a correction was necessary. The following procedure was used for 1 E+0 correcting the change in refractive indices due to addition of water. A refractive index n' was computed using the following: 1E-1 ........ i ....... 0." 1.0 10.0 1.585do3 + 1.33(d'3- _ do3) Diameter (l.u'n)
Figure l c Size distributions from PCASP and FSSP-300 for flight 25b (RH 86%)
n'=
d~
(1)
Liu et al.
532 1.8
r
]
,
1
'
I
'
I
T,'r
~
'
1.6
"~1761.4
1.2
.
o,~,r o
o
Then n', d and the Mie response curves for the FSSP-300 over a range of refractive indices (1.3-1.6) were used to obtain a new wet diameter d'. This new d' were used in a second iteration to calculate a new n". Two iterations was found to be sufficient for satisfactory convergence. The d/do vs the RH plot of the data corrected for the refracted index are shown in Figure 3. The effect of the correction is to increase the values of d/do. The curve is a 4 th order polynomial fit to the data points.
9 o 9 o:
o
DISCUSSION Two models are used here to simulate the d/do versus RH 100 0 2o 40 6o 8o and compare with the d/d0 from Fig. 3. The models are Relative Humidity (%) based on the measured chemical data and the particle Figure 2 d/do for dry diameter of 0.4 l.tm growth curves of Tang and Munkelwitz (1977). The first as a function relative humidity model assumes that only species comprised of H +, NH4 + and 8 0 4 = governs the water absorption by the aerosol. The , second model assumes that all the measured soluble 1.8 .... ~-- I ....' . . . . . I ' I ' species including the organic ions controll the absorption of water, and that their water activity can be characterized 1.6 by that of NH4HSO4. The values of d were computed for each case using the following equation: "~1761.4 1.0
_.__L__
I
,
I
,
I
,
I
'
l
9
d= 1.2
O
9
+ [do'* 0 -
(2)
sol
(19
9
where eson is the soluble volume fraction based on the ion concentrations relative to the volume computed from the 1.0 100 PCASP measurements (taken from Li et al., 1996), and 2O 4O 6O 80 0 Relative Humidity (%) (d/d0)son is obtained from the theoretical growth curves at Figure 3 Refractive index corrected d/do the measured RH. In the first model ~ol is assumed to be as a function of relative humidity determined only from the volumes of NH4 § and SO4- in the samples, and the ( d / d o ) s o l is based on either NH4HSO4 or (NH4)2SO4 depending on the stoichiometry of the sample (21 samples were represented by (NH4)2SO4 and 14 by NH4HSO4). In the second model ~oi is assumed to be determined from the sum of the inorganic and organic ions (also from Li et al.), and the (d/do)sol is based on NH4HSO4. The d/do calculated from the first model are plotted in Fig. 4. The polynomial fit to the data of Fig. 3 is also shown. With few exceptions, the calculated d/do fall significantly below the data fit. In part, this is because the simulated d/do does not take into account the hysteresis of (NH4)2804, but it is evident that more than half of the points for which this is not a factor (i.e. d/d0 is > 1) also fall below the data curve fit. The results of the d/d0 simulation from the second model are shown in Figure 5. Again the polynomial fit to the data in Fig. 3 is included. In this case the simulated d/do appear to follow the data fit better, however, at higher RH the d/do are generally above the data curve and at lower RH they are generally below the curve. Here again, the simulated d/do does not take into account the hysteresis of NH4HSO4. Also, the (d/do)sol is based on data collected at 25~ and the median temperature of data collection is about 17~ (range: -7 ~ 28~ The fact that the d/d0 exceed 1 at lower relative humidities in Fig. 3 may be due to one of two possible reasons. The first is that the probe measurements did not overlap for dry conditions, and that the d/do at lower RH represents an offset in the probe measurements. Although the PCASP and FSSP_K
I
,
I
i
I
,
I
Comparisons between hydrated and dehydrated aerosol particle size distributions
533
300 were calibrated together with similar particles, it is possible that changes occurred during the measurement program that resulted in such an offset, ff this were indeed 1.6 an instrument issue then all the d/do in Fig. 3, and the curve fit, would have to be shifted downward by about 0.1; not necessarily constant with respect to RH, due to finite 1.4 bin sizes. Such a shift would favor model 1 relative to 2. "0 9 9 The second possibility is that the droplets were in a metastable state at the lower RH, which would favor 1.2 model 2. If this is true then the implication is that NH4 + ~ 9 and SOn- do not account for all the growth of the particles seen in these data, and that the organic ions are responsible 1.0 20 40 60 80 0 100 for much of the water uptake. This is consistent with the Relative Humidity (%) results of Saxena et.al. (1995) who have discussed the Figure 4 d/do from model 1. The curve is hygroscopic behavior of carboxylic and dicarboxylic acids, the data fit from Figure 3. among other organic species, which can be similar to those of sulfates. These preliminary results tend to agree with previous studies highlighting the importance of organics to the uptake of water by the atmospheric aerosol. It is hoped that a more detailed comparison of the two probes used here will permit a better resolution of the water uptake, and more cases will be examined in order to gain a representation of the water component of the aerosol as a function of altitude. 1.8
'
1.8
r
1.6 o "0
1.4
'
I
'
I
'
1
.
REFERENCE Li S-M., Banic C.M., Leaitch W.R., Liu P.S.K., Zhou X.L., and Lee Y.N. (1996) The water soluble fractions of the aerosol and their relation to the size distribution based on aircraft measurements during NARE 1993. J. Geophys. Res., 1996 (in press). Saxena P., Hildemann L.M., McMurry P.H. and Seinfeld 9 J.H. (1995) Organics alter hygroscopic behavior of atmospheric particles. J of Geophys. Res., 100, 1875518770. Strapp J. W., Leaitch W.R. and Liu P.S.K. (1993) Oo , Hydrated and dried aerosol-size-distribution measurements 100 from the Particle Measuring Systems FSSP-300 probe and 20 40 60 80 the deiced PCASP-100X probe. Atmos. and Oceanic Relative Humidity (%) I
'
1
'
I
'
I
"7.,
1.2
1.0
I
0
Figure 5 d/do from model 2. The curve is Tech., 9, 548-555. the data fit from Figure 3 Tang I.N. and Munkeleitz H.R. (1977) Aerosol growth studies-Ill. Ammonium bisulfate aerosols in the atmosphere. J. ofAerosol Sci., 8, 321-333.
PROPERTIES
OF RADIOACTIVE
AEROSOLS
IN THE ATMOSPHERE
G. LUJANIENE, V. LUJANAS
Institute of Physics, A Gostauto 12, 2600 Vilnius, Lithuania B.I. OGORODNIKOV, A.K. BUDYKA, V.I. SKITOVICH Karpov Institute of Physical Chemistry, Obucha 10, Moscow, Russia Abstract - Aerosol samples were taken in Vilnius and Preila (Lithuania) and in the middle
troposphere (at the altitude of up to 6 km). Physico-chemical forms, activity median aerodynamic diameter (AMAD) of 134'137Cs, 9~ 7Be, 32p, 33p, 35S and mass median aerodynamic diameter (MMAD) of stable S and P carriers and their changes in the atmosphere were investigated. It has been determined that depending on the season and the sampling place the properties of radionuclide carriers changes. Keywords - 134'137Cs, 9~
7Be, 32p, 33p, 35S, physical and chemical forms
INTRODUCTION The behavior of radioactive nuclides in the environment is determined to a great extent by physicochemical properties of their aerosol carriers. The character of exchange between different natural reservoirs, the rates of atmospheric self-cleaning, and other migration parameters depend on the type of the carrier compound, its solubility and aerosol size. The physico-chemical properties of radionuclide aerosol carriers in the environment depend on many factors. The properties of artificial radionuclide carriers greatly depend on the type of source However, under the influence of various natural factors the radionuclide physico-chemical forms are able to change and at the same time their biological availability changes as well. Cosmogenic radionuclides are formed during the interaction of cosmic rays with the nuclei of air atoms. Aerosol particles are generated from gases through the formation of molecular aggregates in the phase transition. Surface condensation and coagulation are the dominant growth processes for the formation of aerosols the physical and chemical properties of which depend on the type of environment in this region of the atmosphere. METHODS Aerosols were sampled on perchlorvinyl filters of the FPP-15 type by pumping high volume of air (up to 400,000 m3). The ground level air samples were collected in the forested areas in the outskirts of Vilnius. Aerosols of higher atmospheric layers were taken during the plane model AN-24 flight. The same filters (1 m 2 - FPP-15 and 2 m 2 three-layer filters) as in the ground level sampling were used, they were exposed in the car attached to the lower part of the plane. Simultaneously the air was pumped through the three-layer filter packet and the FPP-15 type filter. Aerodynamic aerosol sizes were measured using the multifilter method. To this end special ultra-thin fiber filter packets were used (Ogorodnikov, 1978; Budyka et al., 1993). With a suitable selection of the filter composition and pumping velocity it is possible to achieve rather precise measurements of particle sizes in the range of 0.1-6 ~tm.
After the exposure, filters were detached and Ge(Li)-spectrometry of each filter layer was carried out. According to radionuclide distribution in the packet layers, the parameters of lognormal particle distribution- active median aerodynamic diameter (AMAD) and geometric standard deviation (~) were 534
Properties of radioactive aerosols in the atmosphere
535
determined. Chemical and radiochemical analyses were carried out as described by us previously (Lujaniene, 1991). The investigation of radionuclides chemical forms was based on Tesser (Tesser et al., 1979) sequential extraction method and on the extraction method of different acid concentrations. The supernatant was separated by filtering it through a 0.45 l-tm membrane (Nuclepore, Dubna). Gamma-emitting radionuclides were determined by high resolution gamma spectrometry using a Ge(Li) detector connected to a multichannel analyzer. Beta-emitting radionuclides were measured using low-background UMF-1500 equipment. Taking into consideration the accuracy of chemical procedures, radiometry and air volume measurements, the relative errors were estimated as follows: 7Be - 12%, 32p _ 10%, 33p _ 22%, 35S - 30%, 9~ 18% and 134'137CsO. 1%) the modified nuclei concentration are raised by factor 7-10. New radiolytic aero-
Size spectra, chemical and condensation activity of aerosols and aeroions
545
sols may play the important role in droplet formation if diameter ones more than 3 j.tn~ or if a current supersaturation more then 1%. Size spectra tbr droplet formed on radiolytic nuclei (Fig.3) at S>1~ are typical for initial stage of cloud condensation, i.e. they are similar to Iognormal distribution. At supersaturation S
\
:3
hours
~ i
'
; i
:
.
,
s hoo ,
,
:!! :
2 I_ -+ ()4
Z 1 E+04
OE+O0
/ 1 O0
1000
particle diameter
10000
/ nm
Figure 3 - Temporal development of NaNO_~ Aerosol size distribution in the chamber. Symbols denote SMPS measurements, lognormal fits are denoted by solid lines. PCS measurements, which cover the upper diameter range, are in preparation to be evaluated. H E T E R O G E N E O U S PROCESS : C O N V E R S I O N OF N:O5 TO HNO3 ON NaNO_~ A E R O S O L The oxidation of NO2 by O~ ix investigated with and without the presence of NaNO3 aerosol. Hygroscopic microparticles like salt aerosols exist as metastabil aqueous droplets in the region between deliquescence and recrystallization (Tang and Munkelwitz (1994)). At room temperature NaNO3 aerosol deliquecses at a relative humidity of 74.5 % and recrystallizes at 30 % r.h.. The experiment without aerosol ix performed at a relative humidity of 50 % and a temperature of 295.4 K. The experimental conditions for the heterogeneous experiment with aerosol are a relative humidity of 6 1 % and a temperature of 293.4 K. Since the NaNO3 aerosol is generated by spraying a salt solution, the airstream which transports the aerosol into the chamber is saturated with water vapor. In both experiments the relative humidity adjusting in the chamber exceeds the recrystallization point, therefore tile particles exist in tile form of solution droplets. In the heterogeneous expeiqment the aerosol stream is released into the chaInber for 2.5 hours after the first reactant NO2 has been introduced. The initial size distribution shows a Count Median Diameter of 186 nm with a Geometric Standard Deviation of about 2 at a nurnber concentration of 2.4 9 104 cm -~. The admittance of the second reactant O3 leads to the formation of N205. and HNO~. Figure 4 shows ihe
Physical characterization of aerosols and heterogeneous reactions
569
temporal development of the nlixing ratios of the f)recursors NO2 and 03, as well as the mixing ratios ()f N:O5 and HNO3 for both experiments. In COIltrasI to the homogeneous experiment without aerosol a very fast conversion of N205 to HNO3 is observed, while the N205 mixing ratio is a factor of 20 smaller. The gas-phase chemist O, is discussed in detail by Mentel et al. (1996). We approximate the heterogeneous process as a gas kinetic collision of N205 with the time dependent aerosol surface " d [N205] / dt - - 1 / 4 3' c SAE- [N20] denotes the uptake coefficient, c is the average molecular velocity and SAE represents the total airborne aerosol surface. The aerosol surface as a function of time is determined from the logI-lormal fits of measured size distributions. Furthermore we assume the formation of t~,vo >oas p h a s e , ttNO~ at each successful collision.
y
N20~_ (?{') + H~O.._ (1) --> _9 HNO~ ( g ) ] h e comparison of measured trace gas decays with model calculations yields an uptake coefficient oi 7 0.()3 for N~O5 on wet NaNO~ aerosol. 0
1
2
3
4
5
{;
7
8
9
10
11
12
0
0.5
05
oo
o o -} ~ . s
1
2
3
4
5
,.o
o
. . . . .
0
1
2
3
4
5
6
7 / h
8
9
10
11
12
9
=~ . . . . . . . . . .
0OOoi
,ot~C-----~
- "
10
11
2
_,__= . . . .
'
205 0
1
2
3
I ] |I'
" "
r
duration
8
0.06 t ~ '
0.5
N205
7
j
~
o
6
4
5
__ 6
duration
7
8
9
10
11
12
/ h
Figure 4 - Mixing ratios of NO2, 03, N205 and HNO3 without aerosol (on the left) and with NaNO~ aerosol (on the right). Symbols denote concentrations measured by FTIR, solid lines represent model calculations REFERENCES Dentener, F.J. and Crutzen. P.J. (1993), Reaction of N205 on Tropospheric Aerosols, J. Geophys. Res.. Vol.98,NO.D4, 7 1 4 9 - 7163 Mentel, Th.F., Bleilebens. D. and Wahner, A. (1996) A study of nighttime Nitrogen Oxidation in a large reaction chamber, accepted by AtnTospheric E~,itonnlelzt Tang, I.N. and Munkelwitz. H.R. (1994) Journal q( (;eOl)hys. Resear('/z. Vol.99,NO.D9. 18801 - 1~8()~
AEROSOL NUMBER CONCENTRATION IN ST.PETERSBURC A.D.YECOROV, A.A.SINKEVYCH, V.D.STEPANENKO Voeikov Main Geophysical Observatory,
Karbyshev str., 7, St.Petersburg, Russia Abstract - A new solution presented,
scattering "a
which
g~ves
a
of
the
lidar
possibility
equation
to
obtain
properties of the aerosols without
priori" assumptions.
The
is
the
traditional
unconventional
solution
applied to data of lidar sounding in St.Petersburg.
is
Keywords - Lidar so~mding, atmospheric aerosols INTRODU@TION
The lidar equation involves two unknown
values:
the
backscattering
coefficient and the extinction coefficient. So it is no simple matter to obtain a reasonable description of the
the aerosols from lidar data. Only the
distance
between
lidar
scattering
location
and
properties
scatterir@
of
vol-]
we could write the systci-n in finite differences thus:
P(/ &
(
)
-
_v,.
- q)x + R,-I,, Z.x_ + q-~p.Ol " - k v
_ _
c
; _ y .
I
( ) ':' PI j" = . q) +. R~,7~.~Z~. . +. qXp.()-(,
k~.
_
1
........-x A(p,.o- .U)
p.~. . /
"
1
- _v
A(p,.ci .I')
P.> ~ / I a )
shows that:
i. The ratio CA/CB ~ I A a B / I s a a . ii. The functions A and B go to constants: A = IA/(IA + I8) and B = I B / ( I A + I s ) . This fact simplifies the asymptotic analysis, for the coefficients in eq. (5) become time independent. iii. The asymptotic values of A and B are independent of the rate constants aA and aB- At the initial stage the function A and B are also constants: A - a A I a / ( a A I a + a B I 8 ) , and B = CtB]B/(CtA]A 4- CtB[B).
Source enhanced condensation in binary mLrtures
621
iv. The composition distribution has a sharp maximunl near the point m0, n0. This means that at large tinles tile composition of condensing particles goes to constant (the mean concentratio~l of A-Jnollomers ill the l)article is [A e_ (23)
IA+IB
independently of the rates of the monomer capture). v. In the case when the condensation coefficients depend on the sum m + n or another linear combination of m and 7z (e.g. total particle mass M = #Am + #Bn) the binary condensation problem is reduced to monocomponent one the kinetics of which is described by the equation:
Otc(M,t) = C~AT(M
-
#A)CAC(M -- #A,L) + a B T ( M - ttu)cBc(M - #B,t)--
(O~ACA + aBcB)7(M)c(M,t)
(24)
The particle content is formed independently of the particle mass. Theretore it is reasonable to assume that the particle composition is binomially distributed around the mean concentration ~(t) of A - m o n o m e r s in the particle:
/t A ~ B ~g m ~rn )9-,n WM(m,n) - M(pA -- #B) Cg (1 - (:
(25)
where WM(m, n) is the probability to find exactly m A-monomers in a particle of mass M, C~ are the binomial coefficients and the summation goes within the interval M/#A < g < M/#B (#A > #B). The mean concentration ~(t) may be found from the global balance. At the moment t the concentration of A - m o n o m e r s consumed by particles for their growth is IAt --CA(t). The total consumed concentration is (IA + I u ) t - CA(t)- cB(t). Hence: IAt-~A(t) -
(:~ + : ~ ) t - ~ ( t )
- c~(t)
(%)
CONCLUSION We showed that rather complicated nonlinear time dependent 2d problem of condensation of binary vapor moixtures still may be (asymptotically) solved by introducing the Green functions. The full analysis became possible after the observation that the G-function is concentrated near a point on the composition ( m , n ) - p l a n e , the trajectory of which defines the time dependence of the particle composition. This method is perspective (and will be applied in the near future) for consideration of the nucleation controlled growth of particles in binary mixtures. REFERENCES Laaksonen, A., TManquer, V., and Oxtoby, W. (1995) Annu. Hey. Phys. Chem., 46, 489-524. Lushnikov, A.A. and Kulmala, M., (1995) Phys. t~ev. E 52, 1558-1668. Stauffer, D., (1976) ,1. Aerosol Sci., 7, 319-333.
THE
MOMENT
METHOD
IN THEORY
N.M. KORTSENSTEIN,
OF BINARY
CONDENSATION
E.W. SAMUIIZ)W
Krzhizhanovsky P o w e r Engineering Institute, Leninsky pr. 19, 117927 M o s c o w , Russia A b s t r a c t - The kinetical equation for drop size and composition distribution function w a s proposed and then system o f the m o m e n t equations for description o f binary condensation kinetic were written. C o m p u t e r p r o g r a m for solve its w a s realised. Kinetic formation o f sulphuric acid-water aerosol was calculated K e y w o r d s - Binary condensation; Distribution function; Kinetical equation; M o m e n t equation NOMENCLATURE c = solution drops composition, dimensionless = composition change rate, s 1 C = average composition change rate, s 1 f - - distribution function, kg 1 m I I = binary nucleation rate, m "3 s "1 k = Boltzmanns constant, J K l M = mass o f molecule, kg N = n u m b e r o f molecules in drop p = pressure, Pa r = drop size, m f = growth rate, m s 1 R = average growth rate, m s 1 T = temperature, K t = time, s cz = sticking coefficient, dimensionless p = density, kg m "3 z = relaxation time, s f~ = m o m e n t o f distribution function, kg l m ~~ Subscripts A = w a t e r molecules B = sulphuric acid molecules i, j = m o m e n t n u m b e r s k = drops s = solution v = acid v a p o u r oo = absence o f condensation 0 = constant drop solution composition Superscripts
i, j = m o m e n t n u m b e r s sol = equilibrium solution pressure * = critical e m b r y o
622
The moment method in theory of binary condensation
623
In the theory of bulk condensation for the one component vapour the kinetic of process can be described by the system of ordinary differential equations relatively to the moments distribution function for drop sizes. The kinetic equations for the drop size r and composition c distribution function f(r, c, t) was suggested in the case of binary condensation df ____0(if) + O ( t"6) = -I5 ( r - r*)5(c- c*) dt & Oc p
(1)
where the C=NBMB/(NAMA),the NB, NA are numbers of molecules of B, A component in drop, the MB, MA is the mass of the B, A molecules, i" is the growth rate, 6 is the rate of composition change, I is binary nucleation rate, p is the density of two-component vapour and drops mixture, r*, c* are size and composition of critical embryo. The kinetic equation (1) was transformed to d ~ . j , i~ ~fri_,cjf drdc + j~ ~6ricJ_,f drdc + I(r,)i(c,)j dt rc l.'C
9
*
(2)
~
where f~,j = ]~]~ricJf drdc
(3)
r ~ c ~
is the moment of distribution function. The expressions for t and 6 was obtained in form r OPs. f=f0-~~c 3Ps 0c
(4)
c=
(5)
+1
where, t o = t A + rB,
9 -C~A'B "/MA'B[pA,B_ PA,Bt,T)] sol ~C rA'B Ps(C,T)V2r~kT
(6)
sol (c,T) is solution equilibfimn pressure of components, is the grow rate in free molecular re,me, PA,B and p~3 is partial pressure of components in gas. The i'0 and 6 are depended only from temperature and drop composition. With help of Eq. (3)-(6) from Eq. 2 it was obtained equations 6cJ dc~rifdr + j~6cJ-I dc~rif dr + I(r*)i(c*)J d~"idt - i]~f~ p s dc]~ri-lfdr-3c, - ~ - c i ~-~-10Ps . c
r=
r*
c*
(7)
r*
From integer-differential Eq. (7) it may be obtained the system of moment equation by change t0(c ) and ~c) to their average values, 1~0, C (for zero approach)
d'Qi'J iRo.Qi_l, j "(~~ 1 0Ps -'dt
-
] I r*)i(c*) j,
7 - ~ c ~'2"i,j- J~-i,j-1 + -p(
i= 0...3, j = 0, 1
(8)
Eq. (8) describes binary condensation process by moment method. The computer program for solving the system of Eq. (8) was realized. Complicated difference from analogy problem in one-component condensation is necessary solve the transcendent equation for determination of the two-component critical embryo composition (Mirabel and Katz, 1974) on every step numerical integration. Kinetic
624
Kortsenstein and Samuilow
formation of the sulphuric acid aerosol was tested with help of proposal method and the computer program. Calculations was realized for the T=355K and concentration of the water vapour equal 16%. The kinetic of the sulphuric acid vapour formation in gas phase was taken into account along with the kinetic of bulk condensation with help of the next equation: dcv dt
-
-
-
Cw -tlx T e
-
-
dCvk dt -
(9)
The first term in that equation models the acid vapour formation in gas phase. In the equation the c~ is the vapour mass concentration; the c,,~( is the vapor mass concentration in the absence of condensation. The second term defines the decrease sulphuric acid vapour in bulk condensation process: dCvk 4 f2oj d~,o dt - 3 ~Ps (f2o.0 +f,~,,) dt
(10)
The binary nucleation rate was determined by Kiang and Stauffer (1973). The relaxation time T is defined the rate reaction constant. Was taken in calculation x = 10 gs (the variant 1), 103 gs (the variant 2) and 105 Its (the variant 3). The numerical drop concentration (nk = pD.o,0)is presented on Fig. 1. It follows from Fig. 1 that increase relaxation time lead to decrease nk. The fall of the nk plot is the result of drop coagulation process. Monodisperse approach is used for description this process.
nk(m-3 ) _
J
1014
\ \
\
1013
\
\
1012 \
1011
var.1
"-, var.3
var.2
'~'L.
1010 i
I
i
I I I II I
10
i
I
I
! I III
I
I
I
I
I I III
10 2
I
10 3
i
i
i i
I llil 10 4
I
I
I I I III
I
i
10 5
I
I I I I II I
10 6
time (#s) Fig. 1.
The moment method in theory of binary condensation
625
Cvk-
10-2 _ _ _ _
10
-3
-_
10-4 _
/
\
/ 10-5-
1 0 -6 _
10-7
// 1/
\
',
,
\
var.2 9
~,
///, var.'l ~i .
\ \
var.3 :
!
%
I
I
I % I / i
I
I
III!11
10
I
I
I
I!!11
l
10 2
~
',!
!
!
! !
I
I
I t I lOll
103
a
I
I llilJl
104
! %
i
i
I i I Jill
105
I
i
Ill
I ill
106
time (gs)
Fig. 2. On Fig. 2 is presented: mass concentration of drops (Ck = Pk/P = (4rr./3)P~'~3,o, continuous line), mass
concentration of sulphuric acid in gas phase (c~, pointed line), mass concentration of sulphuric acid in drop (c,4, =ckD.o,~/(D.o,o+D-o.~), dashed line). It follows from Fig. 2 that increase relaxation time lead to increase of Ck and C~k. REFERENCES
Mirabel P. and Katz J. L. (1974) J. Chem. Phys. 60, 1138. Kiang C. S. and Stauffer D. (1973) Faraday Symp. 7, 26.
VARIATION
OF AEROSOL
IN T H E N E A R - G R O U N D IN THE PRESENCE
DISTRIBUTION
LAYER OF THE ATMOSPHERE OF ATMOSPHERIC
FRONTS
V.G.ARSHINOVA, B.D.BELAN, T.M.RASSKAZCHIKOVA, T.K.SKLIADNEVA
Institute of Atmospheric Optics, SB RAS 1, Akademicheskii Ave., Tomsk, Russia Abstract-Variation of the aerosol number concentration and its disperse composition in the presence of atmospheric fronts, as it follows from the results obtained during 19931994 over Tomsk, is discussed. Some properties of aerosol disperse composition dynamics within the frontal zones and certain differences depending on the geographic type of air mass divided by fronts are revealed.
Keywords-Front passage; Number concentration; Variation
Based on the data of hourly measerement obtained during 1993-1994 yrs. at TOR-station in Tomsk (Arshinov (1994)) the variation of aersol number concentration and its disperse composition in the presence of atmospheric fronts has been investigated. Analysis of fronts has been made by means of surface synoptic maps and contour maps of isobaric surfaces. We have analysed 305 events which can be divided into the following
groups
by their types and subtypes
(Vorobiev(1991)
and
Khromov(1948)): -cold fronts: arctic-62, polar-32; -warm fronts: arctic-63, polar-32; -occlusion fronts: arctic-5, polar-33, tropical-11; -surface cold fronts-51; -upper warm fronts- 14. Since we use in this paper only data obtained at a single point and because of a pronounced annual variation of the aerosol number concentration we have used the following data processing technique. For each event of a frontal passage we have selected five hourly values of the aerosol number concentration before the front, one on it, and five after the front passage. To exclude the influence of
626
Variation of aerosol distribution in the near-ground layer of the atmosphere
627
seasonal and diurnal behaviour all the eleven values have been normalized by the front line value. Thus, all data presented in this paper are in relative units. Ni/N o
- -
1,20
^
1,15t /
- cold
( 93 )
....
warm (96)
....
occl.
(49)
"/V',
1,10
1,05
1,00
0,95
!
0,90 .....
L
9
L
. . . . . . . . . .
before fr.
J
9
L
.......
J
after ~.
Fig.1 Figure 1 shows the aerosol number concentration behaviour in the presence of fronts irregardless of their geographical type. Minimum of the aerosol number concentration on the aerosol size distribution curve 1-2 hours before the cold front passage can be a result of most frequently occuring cold front of the second type in which precipitation falls as well as in front of the front line and behind it what causes the aerosol washout. Significant maximum in aerosol number concentration behaviour due to the warm front passage is most likely determined by wind strenthening in the prefrontal zone after which the precipitation area follows thus causing aerosol washout. The aerosol size distribution behaviour in the presence of an occlusion front confirms the assumption on the aerosol number concentration increase due to the prefrontal wind strengthening followed by an extensive precipitation area which causes the aerosol washout started before the front passage and remaining after it for a long time. Analysis of the aerosol number concentration variation for different particle size (Fig.2) shows that 0.41am particles are less influenced by the fronts than lager particles. Under the front passage the relative variations of 0.4gm particles are within 0.9-1.1 range and within 0.9-13.5 range for coarsemode particles (d=2.0~tm).
Arshinova et al.
628
Besides, the shape of number concentration curves for 0.41.tm particles is similar to the integrated number concentration curve (from 0.4 to 10.0t.tm), and it differs from such for other fractions. !,15 i
A
a)
~
1,7[
i
/ /
"J "
0,95i
o.9o.,,
1,2! 5
/''~'
,
t,
i.oi d=0.9
d=0.4
c)
3,0 r
2,5
2,0[
ii
12 I
~i
10 !,
l
,,sL / i/
d)
14
/
//t
..... cold - - warm
- occ/
6
J"Y 4i , d = 1.5
d = 2.0
Fig.2 Minimum at all curves of midle-coarse particles (d=l.01.tm) has been observed during the front passage, and after the passage the aerosol number concentration increases. The prefrontal maximum of the aerosol number concentration can be explained to be the result of wind strengthening. The presence of the second minimum behind the warm and occlusion fronts is still to be interpreted. Curves for the coarse-mode particles have the classical shape. An increase in the aerosol number concentration is observed due to the wind strengthening, which then is followed by the aerosol washout in the frontal zone. Behind the front the regeneration of the aerosol concentration up to its value in a new air mass is observed.
REFERENCES Arshinov, M.Yu. et al. (1994) Atmospheric and Oceanic Optics.Vol.7 N8, 580-584. Vorobiev, V.I. (1991) Sinopticheskaia Meteorologiia. Gidrometeoizdat, Leningrad. Khromov, S.I. (1948) Osnovy Sinopticheskoi Meteorologii.Gidrometeoizdat, Leningrad.
STUDY OF THE ATMOSPHERIC
ULTRAFINE
AEROSOL
M.YU.ARSHINOV, B.D.BELAN Institute of Atmospheric Optics, SB RAS 1, Akademicheskii Ave., 634055, Tomsk, Russia. Abstract-Ultrafine particles are formed in situ from the gas phase by homogeneous nucleation and make the basis for production of submicron aerosols, which have the longest life-time in the atmosphere. Because measurements of the ultrafine particles are very difficult the information about their behaviour is very limited. This paper presents some measurement results on number concentration of such particles obtained during the April 1995 over Tomsk at the TOR-station. Keywords-Ultrafine particles; Size distribution; Diurnal behaviour As known, ultrafine particles are formed directly in the atmosphere (in situ). New aerosol forming substances (AFS) are produced as a result of photochemical and catalytic reactions between some atmospheric gases. The vapour saturation pressure for these substances is lower than that of the initial ones. Therefore, condensation of supersaturated vapour of AFS occurs and as a result ultrafine particles are formed, which then can reinteract with gases (Georgievskii (1986)). All oxygenous gases can generate clusters (Orlova (1983)). Process of AFS vapours conversion into the disperse phase can be derived as a result of homogeneous condensation of the same substance vapour, homogeneous heteromolecular condensation of several substance vapours; adsorption of molecules onto clusters; heterogeneous condensation. According to Rosenberg (1983), clusters are formed as a result of hot noneqilibrium nucleation of carbon, silicates, and some metals upon fuel combustion and mainly, by cold eqilibrium nucleation of water vapour and AFS vapour. Moreover the moisture condensation makes the primary contribution into the cluster generation. After the nucleation the transient grovgth occurs due to the condensation of AFS vapour. Coagulation plays the minor part in the size growth process. Just in this way ultrafine particles are formed. Large portion of these particles lies in the 2 to 200 nm diameter range. This is quite a massive formation which consists of tens and hundreds of molecules. The primary process of ultrafine particles removal is their adhesion to lager ones. Only a small part of ultrafine aerosols reaches the size 80 to 200 nm. Since coagulation of particles of one and the same size is not the primary mechanism of the ultrafine particles removal, then at the consideration of ultrafine aerosols balance Rosenberg derived the following expression describing the typical time of ultrafine particles adhesion to submicron particles: (z-oj)uyp = a~yp/(kT/3rl)la,,,pNsm p , where k is Boltzmann constant; auyp, a sm; are the ultrafine and submicron particles radii, respectively; Nsm p is the number concentration of submicron particles; l is the gas molecule free path; r/is the air
viscosity. For the description of the aerosol number concentration balance he used the following expression: c~
~
=
ln~
ada
= - R a -2,
where a is the rate of the clusters growth due to the heterogeneous condensation of the AFS vapour; R is the typical distribution scale proportional to l.
629
630
Arshinov and Belan
Rosenberg has integrated the latter expression and derived the steady-state distribution of the ultrafine particles: cTN/3a : cons, exp[Rla] .
Several methods to obtain the size distribution of ultrafine aerosol particles can be found in literature. David Y.H.Pui and Benjamin Y.H.Liu (1988) discussed different measuring instruments for the ultrafine aerosol investigations. These are a diffusion battery, a condensation nuclei counter, a differential mobility particle sizer, an electrical aerosol sizer, and scanning and transmission electron microscopes. In our research we use an 8-channel diffusion battery with a synthetic screen as the diffusion collectors designed by Dr.R.A.Mavliev et a/.(1984). It enables us to determine the size distribution of of ultrafine aerosol in the 3 to 200 nm diameter range. In April 1995 we have conducted hourly measurements of ultrafine aerosol particles. Figure 1 shows the number concentration of ultrafine particles. Figure l a presents daily variation of the number concentration integrated over the whole 3 to 200 nm diameter range of ultrafine aerosols obtained with the diffusion battery. All values of number concentration have been averaged to exclude the influence of the anthropogenic sources such as transportation. They represent the average daily number concentration for each hour during the April 1995. As is clear from Fig.la, the integrated number concentration have increased from the sunrise till 11 a.m., whereupon it keeps that value until 8 p.m. From 8 p.m. till 11 p.m. it slowly decreases and after 11 p.m. rapidly falls off down to the night values.
Nem'~4000~
'
35ool
'
~...... ~ ........ " "-a)-- Ncm'~ i
:
i
5oo ~
i
-
200 '- - -
o
'. - - ~ : J -~
......
'
, b)
~
d) --i-
!, i
[ 4o'--
r
:
-
;
-
, . . . .
! !
:
. . . . . . . . . . .
~
Fig. 1 Diurnal behaviour of the number concentration (N) of ultrafine aerosol particles in Tomsk, April 1995, where a) d=3-200 nm, b) d= 10-12 nm, c) d = 16-20 nm, d) d=80-101 nm. Figure 1b presents the diumal behaviour of the smallest particles number concentration. As we have explained above the clusters are formed as a result of photochemical reactions in the atmosphere and they are the basis for the ultrafine aerosol production, therfore the curve 1a has the shape similar to that of the diumal behaviour of the solar radiation intensity with the maximum near 3 p.m. Figure l c shows the daily variation of the aerosol concentration in the 16 to 20 nm size range. This curve has rather different shape than the latter one. The primary maximum of the number concentration of such particles has been observed at 7 p.m. It could be effected by the growth of smaller particles the
Study of the atmospheric ultrafine aerosol
631
generation of which is reduced by this time period. We shall consider this phenomenon later. In Fig. ld the diurnal behaviour of the 80 to 100 nm particles number concentration is presented. It is well seen that there is no significant change in the concentration and it is low during the day. This fact proves the Rosenberg's statement that only a small portion of ultrafine particles reaches the 80-200 nm in diameter. Size distributions of ultrafine aerosol particles obtained at different time over Tomsk are presented in Fig.2. The comparison of Figs.2a and 2b can prove our assumption on the evening maximum in the content of 16 to 20-nm particles.
Fig.2 Size distribution of ultrafine aerosol particles obtained with a diffusion battary at different time, where a) 7a.m., b) 3p.m., c) 9p.m. In conclussion we should like to note that since the begining of January 1996 we measure size distribution of ultrafine particles in a routine mode at TOR-station. We hope to accumulate sufficient amount of data to continue our investigations in this field. REFERENCES Georgievskii,Yu.S. et al. (1986) Optika Atmosfery i Aerosol. Nauka, Moskva,30-42.. Mavliev, R.A. et al. (1984) Kolloidnyi ZhurnaI. Tom XLVI,N6, 1136-1141. Orlova,N.V. (1983) Rukopis dep. v VINITI N2560-83Dep. Pui,David Y.Ho and Liu,Benjamin Y.H. (1988) Physica Scripta.37,252-269. Rosenberg,G.V. (1983) lzv.AN SSSR.FAO.Tom 9, N3, 21-35.
CHEMICAL
AND PHYSICAL
PROPERTIES
OVER FORMER
OF ATMOSPHERIC
AEROSOLS
SOVIET UNION
B.D.Belan, G.N.Tolmachev
Institute of Atmospheric Optics
Abstracts - This paper summarizes the results obtained during 12- Years airborne sounding of aerosol over the former soviet union, with the aircraft-laboratory I1-14 and An-30 "Optic-E"
Keywords - Chemical composition, disperse composition
During the period from 1981 till 1994 the airborne sounding was carried out almost in all geographic and administrative regions over the former soviet union using the aircrati-laboratories I1-14 and An-30. Instrumentation of the aircraft-laboratory has been described in [1]. The aerosol sampling was made using make to filters AFA-XA, AFA-XP and AFA-VP. The isokinetic conditions of sampling were provided by special construction of samples [ 1]. More than 3700 air samples were collected during this period. They embrase air from 200 m to 7500 m. Each sample was analyzed for 27 elements and 12 ions. The detection limits were 20 ng/filters for the elements and 40-200 ng/filters for ions. The disperse composition was made using the photoelectric counter AZ-5. As known [2], the disperse and chemical compositions of atmospheric aerosols are the joint result of many atmospheric processes. Analysis of chemical composition of many aerosol samples as well as of the aerosol particle size distribution has shown that regional processes are among most important ones which determine the chemical and disperse composition of the aerosol. As an example Fig. 1 presents the data collected over most clean region of the FSU (Kamchatka) and in a heavily polluted atmosphere over Kemerovo. The data obtained over West Siberia represent the condition most frequently occurring there. Table one gives some data of a comparatove analysis of the chemical composition of aerosol over different regions of the former USSR. One can easily see distruct regional features from these data.
632
Chemical and physical properties of aerosols over former Soviet Union
103t "~,ClTf3ltlI11-1
10~
3
l~ k ~ 102-
-,,,
~
~2
10-3~
10-4~ I
10-50.5
1 ~,,,I
I
I
1.0 1.5 2.0
I
4
I
I
7 10 --d,/.tm
FIG. 1.Particle size distribution of aerosol over region: 1 - Kamchatka, 2 - West Siberia,3 - Kemerovo city.
633
Belan and Tolmachev
634
TABLE I Chemical composition of aerosol particles over different regions of the USSR, ~ g / m 3 Region i Kamchatka I West Siberia i Kazakhstan [Nizhncvartovsk I Baikal Lake Ulan-Ude
i
Na' K' Ca 2~ CINH ~
0.410 0.570 0.110 0.980 0.350
0.340 0.090 . 3.01 0.220
S()~-
' '-"
400
20 '-""
200
10 i
0 1965
. . . .
! 1970
r
.
.
.
.
.
i
. 975
.
.
.
.
.
. 1980
i
.
.
.
.
1 1985
1990
o9
CO
. . . .
0 1995
Figure 1. Air concentrations of 21~ and "smoke shade" in Southern-Finland (Nurmij~irvi), and TSPconcentrations (Total Suspended Particulate) in downtown of Helsinki during last decades.
AIR QUALITY CLIMATE PERIODICITY AND ITS FUNDAMENTAL REASONS The periodic climate pendulating between continental and maritime, illustrated in Figure 2 with the aid of winter concentrations of particle bound sulphur, is closely related to sea-surface conditions in the Nordic Seas. An accretion fresh water in the North Sea and the Norwegian Sea area, as for example from 1976 to 1978 in connection with the "Great Salinity Anomaly" (Dickson et al., 1988), and at present, since 1995, means a colder sea surface and a more continental climate over Northern Europe. The rather slow circulation of low salinity surface water, 500-1000 km/y, also introduces a long time constant in the climate variation seen in Figure 2. Rahmstorf (1995) has shown, that an intrusion of too much fresh water into the North Atlantic (run-off, melt water, precipitation), may weaken the northern part of the Gulf Stream to a point of no return, so that this part of the "Great Conveyor" would never reach North European latitudes. If the climate warming, which will most likely enhance both melting and precipitation, is due to the greenhouse effect, then Northern Europe will face a drastic change (Broecker, 1995). But if the periodicity, including the present warming, is connected with the variable sun, there is still hope. As stated above, the length of the Northern Europe air quality climatic periods is close to the sunspot cycle (ca. 11 years), but any relation between the two may be partly obscured by the complex connections between sea-surface temperature, wind driven flows of water, and thermohaline circulation in the North Atlantic. It is tempting to speculate whether a point of contact at which a synchronizing force could act can be found between these two cycles.
Analysis of long-term air quality trends and variations in Northern Europe 5
.................... "
[ / [
.......................................
Z " " ~ ,
'".................. " ..................... 5 ! ................. i . . . . . . . . . . . . . . 7_. . . . . . . . "_'. . . . . . " ...................... 5 ............... i .......................
....................... Air concentration of particle-bound sulphur (Dec - Mar, Southern Finland) ~ A r e a l coverage of "heat" (T > 6.5~ at 50 - 200 m depth in he North Sea
697 0
!
c-
-5,, g g $
//::iiiiii,
,
Go ~5 ^~ 40 ~
3
r'~
-9
/:iii: / i~'~{i
2
...
7;
,%{
!V//.I/; ,
i
.....
60 ~, ~ &o4 9
o
s o~
,
1
0 1968
80
1972
1976
1980
1984
1988
-
ii ~ 100
Figure 2. Heat content of the North Sea (inverted scale!) according to Svendsen and Magnusson (1992), and sulphate air concentration in Southern Finland. Numerous attempts have been made to find mechanisms connecting the higher (mainly UV and X-ray) activities at sunspot maxima with weather changes but, so far, it has proved to be too difficult to explain effects on the troposphere because UV and X-rays have already been absorbed in the upper parts of the stratified atmosphere. There is, however, another approach to this question: instead of being hypnotised by the anomalous features at sunspot maxima, we could examine the period between the maxima. After a sunspot maximum, severe magnetic storms are common. Statistically significant evidence indicates that connections exist linking variable solar corpuscular and electromagnetic radiation with the lower stratospheric and upper tropospheric circulation at high latitudes. The time lag between the first appearance of sudden solar activity and its first lower-atmospheric response is 2...6 days, the effects being most pronounced in winter (Roberts and Olson, 1973). Further studies and a promising suggestion of the mechanism, involving polarized ion-induced ice nucleation, has been presented by Tinsley and Heelis (1993). The atmospheric responses include a growing number of severe cyclons with outbreaks of cold air and enhanced cloudiness. But high ionisation rates also occur during sunspot minima, when the effect of the heliosphere shield against galactic and extragalactic radiation is minimal. During these solar minimum years, ozone-destructive nitric oxide production is also higher. These effects are pronounced at high magnetic latitudes. We now have two cyclic phenomena with periods of the same order of magnitude, both possessing sensitive triggering mechanisms, and both being significant wintertimes at high latitudes. Are they linked? When warm saline Atlantic water comes into contact with the arctic air, conditions are ideal for a significant Tinsleyan ion-induced ice nucleation enhancement. Here, if anywhere, is the point at which the triggering mechanisms of the two cycles might tend to synchronise the who periods. The result would be that, from time to time, the general weather mode may include periods tuned to the sun cycle - - that is to say, cloudier and more cyclonic weather and maritime climate with a cleaner quality of air in Northern Europe after sunspot maxima. EPILOGUE This study started as a project searching for fundamental reasons of long-term periodic air quality variations. Unavoidably, however, the analysis met with the question: is it the greenhouse effect or the
Mattsson et al.
698
variable sun that a c c o u n t s for the present w a r m i n g o f the climate? It s e e m s , that if it is the g r e e n h o u s e effect, then the Northern E u r o p e c l i m a t e will face a drastic change. But if, on the other hand, the w a r m i n g is at least partly c o n n e c t e d to the variable sun, there is still hope. 800
t
t
........................................................
r
65
-~Winter (Dec - Mar) means of Pb-210 air concentration in Nurmij~rvi 4 - T o t a l cloud cover in Reykjavik, Iceland E 1:7" m 600
. . . -%
.
.
.
.
.
.
.
.
.
.
.
.
.
.
'
.
.
.
.
'
.
~.
70
J\
o~ v,
i
>
E Z .c:
rr
-
4 0 0
75
.c
r
t.--
._o .,..,
> 0 0
t~ t--(1)
o
co
0 200 .,,.-
80 i i i i i i
i
i
0 1966
,
,
,
,
'
1970
'
'
I
1974
Figure 3. W i n t e r air c o n c e n t r a t i o n s
i '
'
'
,
1978
o f 21~
,
,
,
1982
. . . .
,
1986
,
,,
,
,
85
1990
in Southern Finland air, and total c l o u d c o v e r
(inverted scale) in R e y k j a v i k ( J 6 n s s o n , 1994). Here, it is evident, that the c l i m a t i c p e r i o d s are not restricted to southern Finland and the North Sea (Fig. 1), but are part o f a larger c l i m a t i c area c o m p r i s i n g at least the N o r t h Atlantic and Northern Europe. REFERENCES Bergthorsson, P. (1994). Global impact of climate changes at Spitzbergen. In: Preprintsfrom 19. Meteorology Meeting in Kristiansand, Norway. C/o Norw. Met. Inst., POB 43, 0313 Oslo, 51 - 56. Broecker, W.S. (1995). Chaotic Climate. Scientific American 273, Nr 5, 44-50. Dewan, E.M. & Shapiro, R. J. (1991). Are sunspot-weather correlations real? Atm. Terr. Phys. 53 Nr 1/2, 171 - 174. Dickson, R.R., Meincke, J., Malmberg, S.A. & Lee, A.J. (1988). The "Great Salinity Anomaly" in the Northern North Atlantic 1968-1982. Proc. Oceanogr. 20, 103 - 151. Friis-Christensen, E. & Lassen, K. (1991). Length of the Solar Cycle: An Indicator of Solar Activity Closely Associated with Climate. Science 254, 698 - 700. J6nsson, T. (1994). Cloud climate of Reykjavik. In: Climate Variations in Europe, Publications of the Academy of Finland 3194, 2:23 - 231. Mattsson, R. (1975). Measurements of 2]~176 and 21~ in urban and rural air in Finland. Finnish Meteorological hlstitute Contributions 81, 82 p. Mattsson R., Hatakka J., Paatero, J. and Reissell, A. (1993). The variations and trends of the particle bound sulphur in ground level air in Finland during the last 30 years. In: Mikkanen, P., H~imeri, K., and Kauppinen, E. (eds.) Report Series in Aerosol Science 23, Finnish Association for Aerosol Research, Helsinki, 159 - 165. Rahmstorf, S. (1995). Bifurcations of the Atlantic thermohaline circulation in response to changes in the hydrological cycle. Nature 378, 145 - 149. Svendsen, E. & Magnusson, A.K. (1992). Climatic variability in the North Sea. ICES Mar. Sci. Symp. 195, 144 - 158. Roberts, W.O. & Olson, R.H. (1973). Geomagnetic storms and wintertime 300-mb trough development in the North PacificNorth America area. J. Atm. Sci. 30, 135 - 140. Tinsley, B.A. & Heelis, R.A. (1993). Correlations of Atmospheric Dynamics With Solar Activity. Evidence for a Connection via the Solar Wind, Atmospheric Electricity, and Cloud Microphysics. J.G.R. 98 No. D6., 10,375 - 10,384.
TEMPERATURE DISCONTINUITY DURING STEADY STATE LIQUID EVAPORATION C. A. WARD AND G. FANG Thermodynamics and Kinetics Laboratory, Department of Mechanical Engineering, University of Toronto, Toronto Canada M5S 3G8 Abstract - The conditions existing at the interface of an evaporating liquid and its vapor have become controversial. We report the results of an experimental study that tries to resolve certain of the controversial points. Keywords - Temperature jump, Anomalous temperature profile, Inverted temperature profile INTRODUCTION An understanding of the conditions existing at the interface of an evaporating liquid and its vapor has not been achieved, even though this process has been investigated by a number of different approaches. In the continuum approach, the standard assumption has been that the temperature is continuous across the liquid-vapor interface, but this assumption is now questioned on both theoretical and experimental grounds. (See e.g., Pao (1971), Cercignani et al. (1985), Bedeaux et al. (1990), and Shankar and Deshpande (1990).) The theoretical basis for its criticism arose from investigations based on both classical kinetic theory and nonequilibrium thermodynamics. These approaches led to the prediction of a temperature discontinuity at the liquid-vapor interface during evaporation or condensation, but they also led to controversy because of the direction and magnitude of the temperature discontinuity that was predicted. All of the investigations based on kinetic theory and on nonequilibrium thermodynamics led to the prediction that at the interface of the evaporating liquid, the temperature in the liquid phase was greater than that in the vapor. At the interface where condensation occurred, the prediction from these theoretical approaches was that there was also a temperature discontinuity, but that in the case of condensation, the temperature in the liquid was less than that in the vapor. When these theoretical approaches were applied to predict the temperature profile in a circumstance where a liquid film was present on each of two parallel plates, and a temperature gradient was applied to the system so that one plate was heated and the other cooled, a surprising result was obtained. It was found that the predicted magnitude of the temperature discontinuity by these approaches (for most liquids) caused the predicted temperature profile in the vapor phase between the plates to be in the opposite direction of the applied temperature gradient! Since such a profile is hard to understand physically, it has been labelled as "anomalous" or "inverted", and whether it would be observed physically has been questioned by Koffman et al. (1984). If it were observed, it would indicate the presence of unexpected new phenomena that would have important applications. If it were found not to exist experimentally, it would raise questions about the validity of the theories that led to its prediction. The only way the issue may be decided is by experi-
699
Ward and Fang.
700
mental investigation. However, only one such investigation has been reported to date, and it did not lead to consistent results. Shankar and Deshpande (1990) reported a study of water, Freon 113 and I
i
i=1
~
Pressure Gauge
,_ "-
Vapor out to a Vacuum Pump "Adiabatic" Boundary Manometer
Thermocouple
/ -
,/
Liquid in from a Syringe Pump
mercury. In their experiments, a temperature discontinuity was not observed with water, but they did detect a small one in the case of Freon 113 and a large one in the case of mercury. Shankar and Deshpande (1990) concluded that, on the basis of their data, the existence of anomalous temperature profiles could not be confirmed or denied. Two of the principal difficulties with their experiments were that: (a) they were not conducted under steady-state conditions, and (b) the rate of evaporation could not be measured during an experiment. We report a series of experiments with water in which these two difficulties are overcome.
EXPERIMENTAL APPARATUS, METHODS AND RESULTS
FIGURE 1. Schematic of Experimental Apparatus The apparatus used in this study is shown schematically in Fig. 1. It consisted of three principal components. 1) A flow system: A syringe pump was used to pump liquid into the bottom of a tube at a carefully controlled rate. At the upper end of the tube was a conical section that was made of glass. The system was operated so a liquid-vapor interface formed in the conical section. The position (height) of the interface could be measured from outside the apparatus with a cathetometer. 2) An Ultra-high vacuum system: When present in the conical section, the liquid-vapor interface could be subjected to a vacuum by adjusting a valve that led to a vacuum pump. During an experiment, the pressure in the vapor could be monitored with a manometer that was near the liquid-vapor interface. The purity of the vapor phase could also be measured with a mass spectrometer. Before the start of an experiment, as a part of the cleaning procedure, the vapor chamber was evacuated to a pressure of 10-4 Pa. 3) A moveable thermocouple: The temperature on the center line of the system could be measured by a thermocouple (0.3 mm diameter) that could be moved from outside the vapor chamber. The position of the thermocouple at the time of a measurement could be adjusted over a range of approximately 23 mm. The position of the thermocouple at the time of a measurement was determined with the cathetometer. The vapor phase and cone were enclosed in a double-walled container. The region between the walls was evacuated. The container provided an approximately adiabatic boundary. Thus, the evaporation was taking place in a vessel that had negligible energy coming into the vessel from outside. The energy for evaporating the liquid was provided by thermal energy transported to the interface from within the liquid and within the vapor.
Temperature discontinuity during steady state liquid evaporation
701
EXPERIMENTAL PROCEDURE AND RESULTS Water for an experiment was prepared by passing it through a de-ionizing column, through a distillation column, through a filtering system and then degassing it. A sample of the prepared water was transferred to the syringe pump from the degassing vessel without exposing it to air. The syringe pump was 30 started and the valve leading to the vacuum pump was adjusted to bring the sys25 tem into steady-state. The criterion for steady-state was that once an apparent 20 steady-state had been reached, the system 15 be allowed to operate in that condition for 0 at least two hours, and if at the end of this G 10 0 period, there were no measurable changes ~ 5 00 9 in the height of the liquid-vapor interface ~, 0 in the cone, nor a measurable pressure ,~ 0 9 change, the system was taken to be oper-5 9 ating in steady-state. Once steady-state 9 had been achieved, the rate at which water -10 O In Vapor at Surface was being pumped into the tube leading to 9 the liquid-vapor interface was the rate at -15 9 In Liquid at Surface which evaporation was occurring. While -20
9 0.0
, 0.5
9
, 1.0
9
, 1.5
9
, 2.0
9 , 2.5
9 3.0
Pressure in the Vapor, kPa FIGURE 2. Measured temperature discontinuity at the interface of an evaporating liquid and its vapor. The measurements were made at each of six different steadystate evaporation fluxes. See Fig. 3 for evaporation fluxes.
the system was operating in steady-state, the temperature was measured with the moveable thermocouple at approximately 0.5 mm intervals from near the bottom of the cone in the liquid phase to well within the vapor phase. Special precautions had to be taken at the liquid-vapor interface.
If the thermocouple touched the interface, surface tension effects would cause the interface to attach to the thermocouple. This would disrupt the steady-state. The procedure adopted was to measure the temperature at approximately 0.5 mm from the interface in each of the two phases. The result of these measurements for each of six different, steady-state evaporation rates are shown in Fig. 2. Note that in each case, the temperature in the liquid was less than that in the vapor. This measured relation between the temperature in the liquid phase and that in the vapor is the opposite of that predicted from kinetic theory and non-equilibrium thermodynamics. (See e.g., Pao (1971), Cercignani et al. (1985), Bedeaux et al. (1990).) It is also opposite of that measured by Shankar and Deshpande (1990) for mercury. In their study of water, no temperature difference at the interface was measured. The measured evaporation flux in each of the six experiments is indicated in Fig. 3. The temperature of the water entering the system was 30.4 + 0.9~ Thus, in these experiments, the rate of evaporation was controlled by the choice of the pressure in the vapor phase. As may be seen from Fig. 3, the lower the pressure in the vapor phase, the higher the evaporation flux. By comparing Figs. 2 and 3, it may be seen that as the evaporation flux was increased (by lowering the pressure in the vapor phase) the larger was the observed temperature discontinuity. Both the temperature in the vapor phase at the interface and in the liquid phase at the interface were lowered as the evaporation flux was increased, but the temperature in the liquid phase was reduced more than that in the vapor.
Ward and Fang
702
DISCUSSION AND CONCLUSION During the evaporation of water, the measurements indicate that as the rate of evaporation is increased, the discontinuity in temperature is increased and at each evaporation rate, the temperature in the liquid is less than that in the vapor. The observed 0.40 relation between the temperature in the liquid and that in the vapor is the opposite 0.35 a of that predicted by kinetic theory or by a nonequilibrium thermodynamics. (See 0.30 a [] e.g., Pao (1971), Cercignani, et al. (1985), e~ Bedeaux, et al. (1990).) Also, since the 0.25 observed relation between the tempera"~ [] ture in the liquid and that in the vapor : when evaporation occurs is the opposite ~, 0.20 of that required for the existence of an = .o anomalous temperature profile, these 0.15 results suggest that an anomalous (or ~,~ inverted) temperature profile will not be ~" 0.10 observed for water. These observations [] are in conflict with the theoretical predic0.05 tions and raise questions about their validity. 0.00
,
0.0
,
0.5
,
,
1.0
,
,
1.5
'
,
2.0
'
,
'
Shankar and Deshpande (1990) also
2.5
3.0 studied the evaporation of water, but they did not observe a temperature discontinuPressure in Vapor, kPa ity; however, they did not measure the rate FIGURE 3. Measured water evaporation flux for water of evaporation. Since our results indicate entering the system (see Fig. 1) at a temperature of 30.4 that the degree of temperature discontinu+ 0.9 and evaporating into a vapor maintained at the ity depends on the evaporation flux, one indicated pressure. possibility is that they were studying small evaporation fluxes. For example, at an evaporation flux of 0.091 g/(sm2), we only observed a temperature discontinuity of 1.5~ This small temperature discontinuity could have been overlooked in their unsteady experiments. However, at an evaporation flux of 0.349 g/(sm2), the temperature discontinuity was 14.2~ The latter seems too large a value to be an experimental error. To understand the reasons for the differences between the temperature discontinuity during evaporation at the interface of water and that of mercury requires further study. Since the theoretical approaches provided by kinetic theory and by nonequilibrium thermodynamics do not yield valid predictions of the temperature profile at the interface of an evaporating liquid, there is no theory available for predicting the rate of evaporation at the present time. REFERENCES Bedeaux, D., Hermans, L.J.E and Ytrehus, T. (1990) Physics A, 169, 263-280. Cercignani, C., Fiszdon, W. and Frezzontti, A. (1985) Phys. Fluids, 28, 3237-3240. Koffman, L.D., Plesset, M.S., and Lees, L. (1984) Phys. Fluids 27, 876-880. Pao, Y.-P. (1971) Phys. Fluids, 14, 306-312. Shankar, EN. and Deshpande, M.D. (1990) Phys. Fluids A, 2, 1030-1038.
IONIC
COMPOSITION
OF ATMOSPHERIC
IN WEST
AEROSOLS
SIBERIA
B.S. S M O L Y A K O V , L.A. PAVLYUK AND
1
V.I. M A K A R O V
Institute of Inorganic Chemistry, 630090, Novosibirsk-90, Russia Institute of Chemical Kinetics and Combustion, 630090, Novosibirsk-90, Russia
1. SAMPLING LOCATION AND PERIODS The aerosols were sampled in Klutchi-village, near Novosibirsk, during August 18 -September 2, 1995, and during November 28 -December 28, 1995. Besides, the sample session was also performed in Isetsk (near Tyumene, 900 km to the West from Novosibirsk), during January 12-26, 1996. 2. EXPERIMENTAL The aerosols were accumulated on polymeric fibre filters (AFA-ChA, Russia), with the volumetric rates of 13 cubic meters per hour for Kluchi samples, and about 8 cubic meters per hour - for Isetsk. The duration of each sampling was 24 hours, the total air volumes were 200 -300 cubic meters for each filter. A quarter of each filter was used for the analysis. Each quarter was put in a polymeric vessel with 5-10 ml of deionizated water, and kept here during 24 hours. Then the solution was analyzed by ion chromatography, and by conductometry. Anions were recorded by ion chromatography, Cations - both by ion chromatography, and by conductometry titration. 3. RESULTS AND DISCUSSION The results are shown in Table 1. The mean values and standard deviations are presented for the following data: - pH;
703
Smolyakovet al.
704
- atmospheric concentrations of the following ions: - Ca 2+, Mg2+, NH4 +, Na +, K +, F-, CI-, NO3,
8042-"
- total concentrations of the above ions; - percentage of ionic content in atmospheric aerosols; Besides, there are the ion-equivalents for the above ions (IE, a concentration of an ion, which is multiplied by its charge, and is divided by its molecular weight ). One may see from Table 1 that the ionic percentage is lower in summer (mean value is about 10%) than in winter (30-40%). This is in agreement with the results obtained in the previous years. The mean values of pH of the samples from the Novosibirsk and the Tyumene regions were rather different. Near Novosibirsk the mean pH-value is 5.5, which is close to the equilibrium value of atmospheric
C02 and
a pure water (pHtheor=5.6). Near Tyumene it was about 4.6 (which gives an additional IE=0.04 of H +, see Table 1). Obviously the sums of IE for cations and anions must be equal. One might see from the Table, that the above sums are approximately equal. Thus, this is an indication that we determined ionic contents rather quantitatively. 8042- is the main anion, which contributes 70-80% to the total anion mass. The ions Ca 2+, Mg 2+, and Na + are the main cations in the Kluchi samples (about 70%). The last data are not typical of the usual continental aerosols, where the sulfate ammonia (thus, cation NH4 + ) dominates. As for marine aerosols, Na + and C I ions are the main species. Thus, we might suggest the additional local sources for an income into atmosphere of Ca 2+, Mg 2§ Na +,
and 8042 in summer. The Caspian, and Kulunda
regions may be such sources. Here are the wide areas with soils, containing Na2SO4 (tenardit). The total emission of this substance is estimated to be 4-7 thousands tons per sq. kilometer per year. Due to a snow cover, in winter the emission of the above mentioned substances into the atmosphere is decreasing. Thus, the amounts of the above mentioned ions in aerosol samples could be decreasing, as ,it may be seen from Table 1.
Ionic composition of atmospheric aerosols in West Siberia Table
CATEGORY
pH
1.
IONIC
Ca 2+ +Mg 2+ NH4 +
COMPOSITION
Na +
K+
l.tg m -3
lag m 3
705
OF AEROSOLS
F
, CI
NO3
SO42-
Ion concen,
g g m "3
lag m -3
lag m "3
Ion fract.
lagm -3 ] l.tg m "3
1995
KLUTCHI Mean Stand. dev.
AUGUST 18 - SEPTEMBER 2
0.6
I0.3
0.4
0.2
0.02
0.25
0.25
2.4
4.3
9.5
0.4
0.4
0.1
0.2
0.2
0.02
0.1
0.1
0.9
1.5
4.9
0.005
0.001
0.03
I 0.015 0.017 Cations = 0.067
S u m of IE
1995
KLUTCHI
Stand. dev.
%
5.5
Ion Equiv.
Mean
lag m -3 lag m -3
5.5 0.4
Ion Equiv. S u m of I E
0.047
= 0.059
NOVEMBER 28 - DECEMBER 28
0.5
0.7
0.2
0.2
0.03
0.1
0.8
3.3
5.9
31.4
0.2
0.3
0.3
0.16
0.02
0.24
0.54
1.06
1.7
18.3
0.032
0.04
0.01
0.005
0.002
0.004
0.01
0.07
Anions
Cations = 0.087
ISETSK
0.007 0.004
Anions
1996
= 0.086
JANUARY 12 - 26
Mean
4.6
0.56
2.35
0.26
0.45
0.07
0.16
1.96
9.85
15.6
41.6
Stand. dev.
0.4
0.19
0.76
0.11
0.24
0.06
0.08
1.09
3.55
4.8
10.2
0.033
0.13
0.01
0.01
0.004
0.004
0.03
0.2
Ion Equiv. S u m of I E
Grant INTAS-93-182
Cations(0.183)+H+(0.04)=0.21
Anions = 0.238
A SIMPLE METHOD FOR ATMOSPHERICAL PARTICLE SAMPLING Wen Luo and Folke Peterson
Division of Heating and Ventilation, Department of Energy Technology, Royal Institute of Technology, 100 44 Stockholm, Sweden Abstract
A simple personal sampler is constructed and tested under laboratory conditions. Such sampler can be used to measure particle in dusty areas and the particle mass concentration collected in the sampler can be obtained by means of Junge's distribution. That can simplify the field sampling process and so obtained information can roughly represent individual exposures. Introduction
One important objective of air sampling is to obtain information on the nature and' magnitude of the potential health hazard resulting from the inhalation of particles. That greatly concerns not only the particle concentration and distribution in the sampled ambient but also the individual exposures. However, to accurately assess the hazard for certain harmful particles people need information on the amount of particle that is deposited at the site of respiratory system. A two-stage personal sampler, see Fig. 1, is designed and constructed for the field sampling. bend
I
Iler holder
to pump
Fig. I. Two-stage personal sampler. The sampler consists of (a) a single bend, (b) a filter and (c) a battery-powered personal pump which draws the air through the sampler and can be worn on a person's belt. In principle, the particles collected in the bend simulates the particles deposited the upper
706
A simple method for atmospherical particle sampling
707
respiratory tract and the particles collected on the filter simulates the particles depositing in the lower respiratory region. The experiment for the two-stage personal sampler was taken under laboratory conditions. Di(2-ethylhexyl) phthalate particles were used to determine the sampling efficiency of the inlet and particle collecting efficiency of the single bend and the particle mass collection characteristics of the personal sampler in a wind tunnel. The sampling efficiency of the inlet of the single bend was tested by means of a reference probe. The particle collecting efficiency of the single bend was obtained by measuring the particle concentrations upstream and downstream of the single bend with a Royco 236 Laser Particle Counter. Both the sampling efficiency of the inlet of the single bend and the collecting efficiency of the single bend were obtained as a function of particle size. The particle masses collected in the bend (mb) and on the filter (my) were obtained by weighing the bend and the filter respectively before and after each test run with a Sartorius electronic balance. Based on the experimental data from the sampler, particle size distribution in the sampled atmosphere and collection of different-size particles in the sampler can be roughly obtained by using Junge's size distribution (Junge, 1952) which has been generally used to express atmospheric and indoor particle size distribution for both natural and artificially produced particles. The general expression is written as:
dN
=Adn
dlog(d) where
N
= number of particles per unit volume,
d
= particle diameter, ~m,
(1)
A, n = constants for a given situation. The constants A and n can be mathematically obtained from the amount of the particle mass collected in the two-stage sampler and the particle size distribution in the sampled air can be assessed. The particle mass concentrations collected by the single bend and the filter as a function of particle size can also be obtained. Particle
mass
collection
The test of the personal bend was undertaken in the wind tunnel. The sampling air velocity through the sampler (Vs) was determined by the personal pump and the air velocity in the wind tunnel (Vm) was adjustable. The particle mass collected in the different parts of the sampler under various sampling conditions is shown in Table 1.
708
L u o a n d Peterson
Table 1. Particle mass collected in the bend and on the filter. Conditions
mb. 10 8, g/cm 3
mf. 10 8, g/cm 3 m. 10 8, g/cm 3
m/m, %
Vm = Vs
4,214
4,214
8,428
50
V m = 0,62Vs
4,643
4,214
8,857
47,7
Vm = 0,3Vs
4,429
4,214
8,643
48,8
Particle distribution Under isokinetic measurement conditions, the sampled mass concentration (ms) will be theoretically equal to the real mass concentration (mm) in the atmosphere, i.e. ms = mm. The sampling errors from both anisokinetic sampling and the angle deviation can be corrected by the following general expression (Vencent, et al. 1987): Ill s mm
where
-~s
0
= angle of the orifice of the sampler to the wind, radians,
O~s
= a dimensionless parameter, given by
O~s ~
j(St, Us, urn, 0 ) = 1-
1 1 + k i Sts
(3)
In the dimensionless parameter, Sts is a modified Stokes number, given by
{
(Vmll'sinl/20 }
Sts = St cosO + 4 (~s/
where
(4)
St = Stokes number, ki
= impaction coefficient = 2,1.
The mass concentration of the particles in the air in the wind tunnel can be obtained from:
m m
-~
ms
1 + Ors Ivm Vs
(cosO)- 1)
(5)
Then, the particle size distributions calculated from mean values in the wind tunnel under various conditions are expressed as a function of aerodynamic diameter. Equation (6) is for Vm =Vs, equation (7) is for Vm = 0,62Vs and equation (8) is for Vm = 0,3Vs.
A simple method for atmospherical particle sampling
709
dN = 8,6. 10.6 da 2'465 dlogda
(6)
dN = 10,24. 10.6 de~2'45 dlogda
(7)
dN = 10,86. 10.6 da T M dlogda
(8)
Mass c o n c e n t r a t i o n s The mass concentrations in the sampled air, collected by the bend and filter can be obtained as well. The data used in the calculations were those obtained from the isokinetic samplings (Vm =vs ). dm 10" 9 dm dm, bend d m , filler
10 " 1 0
~t"
o
lo I o 9
,o ~
10-1
1
. . . . . . . .
lO -1
i
9
I
| | |
,o -..
9
9 =" '-
10 0
Aerodynamic
9- 101
diameter,
~m
Fig.2. Particle mass concentration in the sampled air and collected by the bend and the filter.
Summary Since the sampler simulates the human respiratory tract, the particle collected in the sampler, then, can be used to estimate the deposition in the respiratory regions; the particles collected by the bend is for the particle deposition in the upper respiratory tract and those collected by the filter for the particle deposition in the lower respiratory region. most of the particles greater than 5 lure in aerodynamic diameters were collected by upper respiratory tract.
References Junge, C." J. of Meteorogy, Vol. 12, pp 13-25. 1955. Vincent, J. H. and Mark, D." American Industrial Hygiene Association Journal. Vol. 48(5), pp 454-457. 1987.
M O N T E CARLO SIMULATION OF C O A G U L A T I O N PROCESSES IN A LOCALLY ISOTROPIC T U R B U L E N T REGIME. K.K.Sabelfeld, A.A.Kolodko and A.I.Levykin C o m p u t i n g C e n t e r , S i b e r i a n D i v i s i o n of R A S , L a v r e n t j e v a p r o s p . 630090, N o v o s i b i r s k , R u s s i a E-mail:
[email protected] ABSTRACT The influence of fluctuations caused by turbulent intermixing on the coagulation process is studied numerically, by Monte Carlo simulations and a finite element method. Two situations are analysed in details: in the first one the intensity of collisions is fluctuating, and in the second the intensity of the monomer generator is a random process. It is shown that both these fluctuations do strongly influence the size distributions. INTRODUCTION It is generally known that there exists an effect of fluctuations of temperature and vapor concentration on the particle formation rate (both by homogeneous and heterogeneous nucleation). A qualitatively estimations were given by Hidy and Friedlander,(1964); see also measurements by O'Konsky (1956,1960), and a study by Brock et al. (1986,1988). A recent review of papers in this field is given in Lesniewski and Friedlander (1995). The first main step in this problem is to evaluate the structure of the fluctuations. In the next step one of the approaches based on the classic Becker and DSring expression or the Frenkel theory for the rate of homogeneous nucleation, or on the kinetic description can be applied. It should be noted that both these stages are far from being well studied. For example, Pompei and Heywood (1972), Lockwood and Naguib (1975), and Villsenor et al. (1992) used a simple Gaussian model for the fluctuations. Lesniewski and Friedlander (1995) used the experimentally measured values of the probability density function (pdf) obtained by Sreenivasan et al. (1977). Both the Gaussian and measured pdf's ignore the spatial and time correlations of the concentration fluctuations. In Zapadinsky et al. (1995) we studied the effect of fluctuations on heterogeneous nucleation where the time correlations of the concentration fluctuations were taken into account via a stochastic model of turbulence suggested by Sabelfeld (1991). However in this study we also assumed that pdf is of a gaussian type which can not be correct, as the measurements and calculations show. We mention also the study of Kasper
710
Monte Carlo simulation of coagulation processes in a turbulent regime
711
(1984) who found that coagulation is faster when the particle concentration is inhomogeneous and a simple model of chemical reactions with random factors proposed by Lushnikov (1989). Note that the conventional approach to describe the coagulation in turbulent regime is based on the Smolouchovsky equation with the relevant coagulation kernel (see Saffman and Turner (1956), and Williams and Loyalka, (1991)). Two weak points of this approach can be mentioned: first, it contains some ad hoc assumptions (e.g., it is assumed that the distribution of the gradient of the fluctuating velocity is Gaussian), and second, it describes only the average particle size spectum. In the present report we develope a Monte Carlo method for solving the Smolouchovsky equation with random coagulation kernel and random source. Mathematical justification can be found in Sabelfeld et al. (1996). The main issues of discussion of the present report are the numerical results. STATEMENT OF THE PROBLEM Let us consider the Smolouchovsky equation with a random kernel 0hi
1
Ot- = -2 E
Kijninj - nt E Ktini + Fz(t)
i+j=l
(1)
i=1
with a deterministic source of/-mers and initial condition hi(0) - n l(0) In (1) nl is the number density of l-mers, 1 = 1, 2,..., the kernel K~j describes the collision and coagulation between the i- and j-mers. Since K~j = ozjf~ij (oij is the cross-section of interaction, ~j is the mean relative velocity between the i- and j-mers), the fluctuations of K# are caused by the fluctuations of the relative velocities. The statistical structure of the relative velocities in a locally isotropic turbulent flows is well studied (see, e.g., Sabelfeld, (1991)). Our calculations based on stochastic models showed that the relative velocities v~j have under some assumptions a log-normal distribution. I n t r o d u c e the notation: I(ij -- (t(ij) + I(~j, ni(t) = (?),i(t)) + n'i(t ), where the angle brackets denote the average over the random fluctuations K~j. Averaging tile equation (1) yields O(nl>
1
1
i+j=l
,
i+j=l
oo
oo
-(nz) E[t(~z)(n,) + (I(;tn',) ] - E[(K,t)(n',n~) + (n,)(I(;~n't) ] - E(t(;zn'~n't) + Fz(t). i=1
i=1
i=l
This equation coincides with the standard Smolouchovsky equation if the functions (n'in~) , (KilnS) , and (K~jn'~n~) all vanish. Of course, in general this can not be true. Therefore, different closure hypotheses can be made to obtain a closed system for the average (n,(t)). For instance, if we suppose t h a t (I(~jn'i) = Dij(ni),
(n'in~) - Bij(ni)(nj),
and (t(:jn~in~)
- O, we conle to the Smolouchovsky
equation (1) with [(~j = (K~j) + 2D,j + B#. Tile range of applicability of these approximations can be verified by Monte Carlo simulations which solves the original stochastic Smolouchovsky equation and calculates the desired correlations.
Sabelfeld et al.
712 /
-}
,.L
I I
I
l
o
l
~
~
'~
l
0.8 n'k
n(0)
w
x
l
o
o o
~,'~o ,
0.6
l
deterministic, solution average solution
~
o
oo
o
oo
OOo
~
~ o
0.4
0.2
0 0
Fig.1
1
1
1
2
4
6
The time dependence
___
1
.
~
1
1
1
_L_
1
10 T
12
14
16
lb
20
of lllOllOlll(~.l c o ] l c e n t r a t i o n , for t h e free I n o l e c u l e collisio~l
r e g i m e a n d tile a v e r a g e m o n o m e r
c o n c e n t r a t i o l l for r a n d o m
1
I
0.1
I421.
I
I
d e t e r m i n i s t i c solution - average solution, a = .1, ,
.3,
~r
le-05
le-lO
le-15
i
5
10
30
50
7{
S Fig.2 T h e t i m e d e p e n d e n c e s of m o n o m e r c o n c e n t r a t i o n , for r a n d o m a n d d e t e r m i n i s t i c sour('es~
Monte Carlo simulation of coagulation processes in a turbulent regime
713
The same considerations are applicable in the case of Smolouchovsky equation with random source Fl(t). In practice, the fluctuating part of the source can often be described as a Gaussian stochastic process.
In our calculations we assumed that Fl(t) = ~1,1[1 + cry(t)], where ~ ( t ) i s a
stationary Gaussian process with zero mean and the exponential correlation function B(~) = (~(0)~(~)) = a ~ x p ( - ~ , ~ ) ,
and a is a constant. This implies that we have a stationary random monomer source whose mean is equal to 1, and the characteristic correlation time is a -~
The kernel Kij was taken as in the
case of free molecular regime. NUNERICAL RESULTS Results of calculations are shown in Figs.l,2. In Fig.1 we present the time dependences for the number of monomers: it was supposed that only the collision coefficient Kll was random (lognormally distributed). The results show the influence of fluctuations on the number of monomers. The calculations showed that for dimers this influence is much less. In Fig.2 the kinetic curve is given for the stochastic monomer source described above. It is seen that even for small intensity fluctuations (a = 0.3) the kinetic curves for the problem with the deterministic source and with the stochastic source (but with the same average intensity) differ for all time instances. When the fluctuations are less than a = 0.1, the difference is neglible. REFERENCES Brock, J.R., Kuhn, P.J. and Zehavi, D. (1986). J.Aerosol Sci. 17,11. Brock, J.R., Kuhn, P.J. and Zehavi, D. (1988). J.Aerosol Sci.19, 413. Hidy, G.M., and Friedlander, S.K. (1964). AIChE J.10, 115. Lockwood, F.C., and Naguib, A.S. (1975). Combust. Flame 24, 109. Villsenor, R., Chen, J.Y., and Pitz, R.W. (1992). AIAA J., 30, 2552. Higuchi, W.I., and O'Konski, C.T. (1960). J. Coll. Sci., 15, 14. O'Konski, C.T. and Higuchi, W.I. (1956). J.Phys. Chem., 60, 1598. Pompei, F. and Heywood, J.B. (1972). Combust. Flame, 19, 407. Sreenivasan, K.R., Antonia, R.A. and Stephenson, S.E. (1977) University of Newcastle (Australia) T.N.F.M. 5. Lesniewski, T., and Friedlander, S.K. (1995) Aerosol science and Technology, 23, 174. Sabelfeld, K.K., (1991). Monte Carlo Methods and applications. Springer-Verlag, Berlin-New York-Heidelberg. Zapadinsky, E.L., Sabelfeld, K.K., Kulmala M., Gorbunov, B.Z., and Rackimgulova, D.M. (1995) Journal of Aerosol Science, 26, 1179 Kasper, G., (1984). J. Colloid Interface Sci., 102, 560. Lushnikov, A.A., (1989). Chemical Physics, 134, 259. Saffman, P.G. and Turner, J.S. (1956). J. Fluid Mech., 1, 16. Williams, M.M.R. and Loyalka, S.K. (1991). Aerosol Science Theory and Practice. Pergamon, Oxford-New York-Seoul-Tokyo. Sabelfeld, K.K., Rogasinsky, S.V., Kolodko, A.A. and Levykin A.I. (1996).
Monte Carlo Methods and Applications, 2, 43.
MEASUREMENTS
OF THE CONTENTS
OF ORGANIC
AND INORGANIC
CARBONS, 3,4-BENZO(A)PYRENE,
AND DIBUTYL PHTHALATE
AEROSOLS
IN ATMOSPHERIC
IN WEST SIBERIA
V.I. MAKAROV, Y.N. SAMSONOV, V.V. KOROLEV AND K.P. KOUTSENOGII Institute of Chemical Kinetics and Combustion, Siberian Branch of Russian Academy of Sciences, 630090, Novosibirsk-90, Russia
Atmospheric aerosols are composed of the complex mixtures of natural and anthropogenic chemical species that include sulfates, soluble inorganic compounds, and seasalts, carbonaceous materials, insoluble solid minerals, and metals. Since the effects of aerosols on climate and environment depend on the physical, chemical, and biological properties of aerosols, the assessment of these effects might include all aerosol species. A carbonaceous aerosol component, is next to sulfate the most significant contributor to aerosol climate and environment forcing because of its abundance, the chemical, toxic, optical, and nucleation properties. Sub-micron carbonaceous aerosols are composed of two classes of materials referred to as black (also known as "elemental" or "graphitic") carbon and organic carbon. Organic carbon is a synonym for the total organic aerosol fraction, which is composed of a large number of individual organic compounds (many of which are known as carcinogens, e.g. 3,4-benzo(a)pyrene). Both organic and black carbons affect the extinction of solar radiation. Black carbon aerosols are the light-absorbing species and therefore may cause the warming of atmosphere. Organic carbon aerosols scatter light and together with sulfate may cause the cooling of atmosphere. Carbonaceous aerosols may be primary and secondary, depending on their sources and formation mechanisms. Black carbon is a primary pollutant produced by incomplete combustion of fossil fuels and different biomasses. Organic carbon may be either primary if introduced directly into the 714
Organic and inorganic carbons in atmospheric aerosols in West Siberia
715
atmosphere in the form of particles by various combustion and natural sources, or secondary if produced in the atmosphere by gas-to-particle conversion of anthropogenic and biogenic precursor gases. The pollens from flowering plants are the sources of bioorganic aerosols too. Consequently, the relative concentrations of the different kinds of carbonaceous aerosols over different regions are variable, depending on the kind and distribution of combustion sources, on seasons, etc.. In this article we focus on the daily and season variations of carbon contents, and of a few other chemical compounds into atmospheric aerosols in West Siberia. The atmospheric aerosols were accumulated on aerosol filters, both on glass fibre papers (Dassel 50 mm, Germany, 1.4 and 1.8 m 3 hour -l, 24 hours of sampling), and on polymeric fibre filters (AFA-ChA20 sq. sm., Russia, 5 and 13 m 3 hour l , 24 hours), continuously during 10 - 30 days in various seasons, on the heights of 2-20 m over landsurface. Samplings were performed simultaneously in Academytown (30 km to the South from Novosibirsk), in Klutchi-village (10 km to the East from Academytown), and near Lake Chuny (300 km to the West from Novosibirsk). Besides, an independent sampling was made in Isetsk-town, 60 km to the South from Tyumene (900 km to the West from Novosibirsk). The total mass concentrations of aerosol impurities were measured by weighting the polymeric filters before and after sampling. A quarter of each polymeric filter was used to extract dibutyl phthalate (DBP), 3-4-benzo(a)pyrene (BP), and other organic compounds (e.g. pesticides), followed by the gaschromatography recording of DBP (and pesticides), by using an electron-capture detector. Then the extractants (n-hexane) were evaporated out of test-tubes, and n-octane was added into the tubes. These samples were used for recording 3,4-benzo(a)pyrene by the method of low-temperature fluorescence of BP in the n-octane frozen glass matrix (77~
the Shpolski method). This method has the extremely
good selectivity and sensitivity to 3,4-benzo(a)pyrene analysis. The method of the measurements of organic and inorganic carbons on a glass filter was based on a catalytic conversion of organic substances and black carbon to methane, followed by its recording by gas-chromatography using a flame-ionization detector.
Makarov et al.
716
Figures 1 and 2 show the daily and season variations in the percentage of "organic plus inorganic carbons"
against the total mass concentrations
of aerosol impurities,
and the variations
in
"organic/inorganic carbons" ratio. The mean total carbon percentages were about 10-12% of the total aerosol masses (Cm were -- 40 gg m-3). Taking into account that there were other atoms (H, O, N, CI) in organic molecules, one could estimate the mean percentage of "organic plus inorganic" substances as 15-20%. In summer there were more organic carbons (the
Corg/Cinorg ratios
were about two), but in
winter the ratios were somewhat less than unity. More black carbons are introduced into atmosphere in winter, due to intensive burning of coal, mazut, and fire-wood in this season.
30
Fig 1
25
A- Klutchi, D ~ 1995 B - Iselsk, Jcnucry 1996 C- Klutohl. AugJst 1995 A
(
CI
o
a~15
006
10
A
,'
r" ._
+
I
o
(9
o
A- Klutohi, Deaamb~ 199 B- Isetsk, Jcnuay 1996 C - Klutchl, AugJst 1995
l-
B o
E 20
Fig 2
ii I]
B
,
/~' 'i
,
o
' d
(2)
O "J
5
I
6
,
I
,
I
,
I
.
I
.
I
,
I
.
I
,
I
i
I
,
I
,
I
.
I
8 10 12 14 16 18 20 22 24 26 28 30
Dde
I
6
,
I
zXi .
I
.
I
.
I
.
I
,
I
.
I
.
v I
.
I
.
I
.
I
,
I
8 10 12 14 16 18 20 22 24 26 28 30
Dde
Organic and inorganic carbons in atmospheric aerosols in West Siberia
717
The examples of daily variations in the concentrations of BP and DBP in atmosphere are shown in
13_ ii- DBP,/L I
~, 3,0
'•
-
'
Fig 3. a
I
'
a c-
I
'
I
'
I
'
BP. Aca:lamy-town: DBP, Ao_x~-ny-tawn;
~
I
'
b d-
I
Fig. 3. The data are presented as the ratios of the '
I
BP, Klutchi; DBP, Klutd~
measured
concentrations
to
the
Maximum
ckeChuny:
Permissible Concentrations (MPC) of the above
2,5
substances (1 ng m -3 for BP, and 100 ng m 3 for
2,0
o
1,5
\it
/
DBP,
in
Russia).
Obviously
the
above
concentrations are correlated with the stages of
i 1,0
,'t d
0,5 i 0,0 O
25
27
29
Jcnucry
31 1995
2
4
6
urbanization.
They
are
relatively
high
in
Academy-town, are lower in Klutchi-village, 8
and are much lower in the low-populated area
F ebr ucr y
near Lake Chuny (DBP, curve e).
MODELLING
THE FORMATION
OF H2SO4- H20 PARTICLES
LHSA PIRJOLA University of Helsinki, Department of Physics P.O. Box 9, Siltavuorenpenger 20 D, FIN-00014, Finland A b s t r a c t - The factors which affect the formation of new sulphuric acid particles in different at-
mospheric conditions are investigated. An atmospheric chemistry gas phase bo:~ model coupled to a three mode integral aerosol dynamics model is used. The simulations show the dependence of the concentration of nuclear mode particles on initial pre-existing particles, the intensity ofUV radiation, the emissions of DMS and the ratio of emissions of hydrocarbons and NOx present in air. Keywords - Formation of atmospheric aerosols, sulphuric acid, dimethylsulphide, UV radiation INTRODUCTION Atmospheric aerosols have important effects on radiative transfer directly by absorbing and scattering incoming radiation or indirectly by acting as cloud condensation nuclei (CCN) (Charlson et al., 1987). Several recent papers deal with the new particle formation in the troposphere by H2SO4 - H20 nucleation (e.g. Hegg et al., 1991; Raes and Van Dingenen, 1992; Lin and Chameides, 1993; Pandis et al., 1994; Kulmala et al., 1995). Most of these papers concem marine boundary layer conditions, in particular CCN formation from dimethylsulphide (DMS) oxidation. Hoppel et al. (1994) and Van Dingenen et al. (1995) have analyzed measured size distributions of marine aerosols. Based on computer simulations, Kulmala et al. (1995) have derived an analytic expression for the atmospheric conditions in which new particles are formed via SO2 oxidation. The formation rate of new particles depends on pre-existing particles, such as their number concentration, size distribution and composition, and the amount of condensable vapours, in this case H2SO4 vapour. The concentration of sulphuric acid depends on the concentrations of OH-radicals, SO2 and in marine conditions also SOs (Lin and Chameides, 1993). The first depends on both the ratio of nonmethane hydrocarbons (HC) and nitrogen oxides (NO~) present in air and the intensity and wavelength of UV radiation. The anthropogenic sulphuric emissions and DMS emissions from the ocean determine the concentration of sulphuric oxides. Based on modelling, this work discusses how the concentration of HzSO4- H20 nucleation particles depends on the UV radiation, the ratio of HC and NO,, emissions, the DMS emissions and the concentration of pre-existing particles in different conditions (urban, marine and rural). THE MODEL AND THE INPUT VALUES The formation of new particles was simulated by a model in which the atmospheric gas phase chemistry model (Pirjola, 1995) and the three mode (nucleation, Aitken and accumulation mode) integral aerosol dynamics model (Kulmala et al., 1995) were coupled together. The chemistry model is a box model, the height of an air parcel being 1000 m, and for each atmospheric species (67 altogether) chemical loss and production, emission and deposition are taken into account. Besides the gas phase chemical reactions, the nucleation and the condensation on to pre-existing particles also affect the concentration of sulphuric acid. The nucleation mode particle concentration is determined by nucleation and coagulation. The pre-existing particles were simulated by two distinct size-distribution modes: the Aitken mode with an initial radius of 10 nm and the accumulation mode with 100 nm. The growth of particles is determined 718
Modelling the formation of H2SO4-O20particles
719
by condensation and coagulation. The set of differential equations was solved using the NAG-library FORTRAN-routine D02DAF. The simulation time was about 24 hours and it started at 0.00 (UTC) on May 4. The temperature varied sinusoidally in 24 h periods between 0 ~ and 10 ~ and reached its maximun at noon. The relative humidity also varied sinusoidally, but in the opposite phase between values 0.8 and 0.5. The latitude was chosen to be 67.77 ~ which corresponds to the Varrio measurement station in northern Finland (Hari et al., 1994). The initial concentration of the two pre-existing modes was changed logarithmically from 10 to 10000 cm 3. The intensity of the UV radiation was increased in 5 % steps up to 25 %. To be able to make comparative analysis of the influence of the emissions of HC, NOx, SO2 and DMS, the emissions were changed according to Table 1. The classification to rural cases (R1,112), urban cases (U1,U2,U3) and marine cases (M1,M2,M3) is somewhat artificial, but generally follows the values given in the literature.
,,
Concentrations in molecule cm3 Emission rates in molecule cm2 sl Initial NO concentr. Initial NO~ concentr. NOz emission rate HC:NO~ (emissions) Initial SOz concentr. SO2 emission rate Initial DMS concentr. DMS emission rate
R1
R2
U1
U2
U3
M1
M2
M3
0.49.10 l~ 0.49.10 l~ 0.49.10 l~ 0.49.10 t~ 0.49.10 l~ 0.49-10 l~ 0.49-10 l~ 0.49-10 l~ 1.97-10 l~ 1.97.10 l~ 1.97.10 l~ 1.97.10 l~ 1.97.10 l~ 1.97.10 l~ 1.97.10 l~ 1.97.10 l~ 2.77- 10 l~ 2.77- 1011 2.77.10 tl 2.77.1011 2.77.10 II 2.77.101~ 2.77.101~ 2.77- 101~ 15:1 1.5:1 2.46- 101~ 2.46.101~ 2.65- 109 2.65.109
1.5:1 1:6 1:6 2.46.10 I~ 2.46- 101~ 9.41. l0 II 3.98.10 l~ 3.98- 101~ 6.5.10 II
0.8:1 2.46.101~ 2.65.109 3.0-107 2.3.109
0.8:1 2.46- 101~ 2.65- 109 2.5- 109 2.8- 109
0.8:1 2.46-101~ 2.65.109 2.0.1010 0.7-101~
Table 1. Classification of different cases. R E S U L T S AND D I S C U S S I O N Figure 1 shows the typical time development of the concentration of newly formed particles in the different cases. The maximum values are reached in 5 to 10 hours from the beginning of the simulation. The higher values are characteristic to the urban cases and new particles are formed quite rapidly, while the lowest values correspond to the rural cases and the formation is lower then. 105 ~.......
,
lo4t~"
-U3
.......
M3~ U2 I
103 !
i o
r
-,
u,l
102
M1J
lo'
i
lo 0 10 -I
R2i
i
o
a~ i ] ,
5
lb
Hour
l's
2'0
i
2~
Fig. 1. Maximum concentration of newly formed particles in different cases. The initial A.itken and accumulation mode concentrations are 10 and 10 except for the case U3 where they are 100 and 100.
Pirjola
720
More than 100 simulations were made in this study, but only a few results can be presented in this paper. The notable observation is that the maximum values of the concentrations of newly formed particles are strongly dependent in each case on the initial concentrations of Aitken and accumulation mode particles such that they clearly decrease with increasing particle concentrations. The obvious explanation of this behaviour is that a greater initial particle concentration causes more condensation. At the maximum the values varied from some particles to about a hundred particles in rural cases, from some hundreds to a thousand in marine cases and from several hundreds to about ten thousand in urban cases. In the rural cases letting the NOx emission grow ten-fold (decreasing HC:NOx emissions to one tenth) resulted in the newly formed particle concentration becomes 20 to "70 times larger. This can be explained as follows: because NOx is the precursor of ozone an increase of NOx produces more ozone by the photochemical reaction, and ozone in turn produces OH-radicals and so the sulphuric acid concentration is enhanced. Table 2 shows the relative maximum concentration of newly formed particles as functions of DMS emissions in marine cases at different Aitken mode concentration. For brevity, only the case of the lowest intensity of UV radiation and the initial accumulation concentration being 10 is presented in this paper. By considering these data graphically we observe that the dependence is practically linear, which is in agreement with the results of Russel et al. (1994), but some additional simulations are needed to prove this definitely.
Aitken
M1
rel. Nma~ M2
I0 I00 I000
1 1 1
1.27 1.33 1.24
M3 3.61 5.601 4.61
Table 2. The relative nucleation mode concentration at different initial Aitken mode concentrations and DMS emissions ( M I 2.3.10 9, M2" 2.8.10 9 and M3 7.0.10 9 molecule cma sl). The initial accumulation mode particle concentration is 10 in each case. General features of all simulations were that the increase of the intensity of UV radiation increases the maximum concentration of newly formed particles while the increase of the pre-existing particles in the Aitken or accumulation mode decreases that concentration as mentioned above. The enhanced UV radiation produces more OH-radicals and thus the concentration of sulphuric acid grows.
U-V 100 1000 10000
1 1 1
1.05*UV 1.05 1.13 1.18
100 1.I*UV 1.08 1.25 1.39
1.15*UV 1.11 1.38 1.64
1.2*UV 1.25"UV 1.13 1.15 1.51 1.92
1.65 2.21
UV
1000 1. I*UV
1.2*UV
1 1
1.56i 1.51
2.34 2.21
Table 3. The relative maximum values of the concentration of newly formed particles such that the concentrations for the lowest intensity of UV radiation are equal to unity. Each line refers to a constant Aitken mode value and each cohurm to a constant accumulation mode value. Table 3 shows the case U3 as an example. We can see that the increasing influence of radiation is stronger when either the initial Aitken mode or the accumulation mode particle concentration increases. However, for the case accumulation mode equal to 1000, an increase in Aitken mode slightly decreases the effect of UV radiation. Because the amount of data is very small, definite conclusions require additional simulations. By plotting the concentrations as functions of UV radiation the dependence appears to be linear. The linear regression fits gave correlation coefficients close to one.
Modelling the formation of HeSO4-H20 particles
721
REFERENCES Charlson, R.J., Lovelock, M.O.A. and Warren, S.G. (1987) Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate. Nature, 326, 655-661 Hari, P., Kulmala, M., Pohja, T., Lahti, T., Siivola, E., Palva, L., Aalto, P., H~.meri, K., Makel~., J., Vesala, T., Luoma, S., Anttila, P. and Pulliainen, E. (1994) Air Pollution in Eastern Lapland: Challenge for environmental measurement station. Silva Fennica, 28(1), 29-39 Hegg, D.A., Radke, L.F. and Hobbs, P.V. (1991) Measurements of Aitken nuclei and cloud condensation nuclei in the marine atmosphere and their relation to the DMS-cloud-climate hypothesis. J. Geophys. Res. 96, D10, 18727-18733 Hoppel, W.A., Frick, G.M., Fitzgerand, J.W. and Larson, R.E. (1994) Marine boundary layer measurements of new particle formation and the effects nonprecipitating clouds have on aerosol size distribution. J. Geophys. Res. 99, 14443-14459 Kulmala, M., Kerminen, V-M. and Laaksonen, A. (1995) Simulations on the effect of sulphuric acid formation on atmospheric aerosol concentrations. Atm. Env. 29, 377-382 Lin, X. and Chameides, W.L. (1993) CCN formation from DMS oxidation without SO2 acting as an intermediate. Geophys. Res. Lett. 20, 7, 579-582 Pandis, S.N., Russell, L.M. and Seinfeld, J.H. (1994) The relationship between DMS flux and CCN concentration in remote marine regions. J. Geophys. Res. 99, D8, 16945-16957 Pirjola, L. (1995) Ph. licenciate thesis (in Finnish), Department of Physics, University of Helsinki Raes, F. and Van Dingenen, R. (1992) Simulations of Condensation and Cloud Condensation Nuclei from biogenic SO2 in the remote marine boundary layer. J. Geophys. Res. 97, D12, 12901-12912 Russel, L.M., Pandis, S..N. and Seinfeld, J.H. (1994) Aerosol production and growth in the marine layer. J. Geophys. Res. 99, D 10, 20989-21003 Van Dingenen, R., Raes, F. and Jensen, N.R. (1995) Evidence for anthropogenic impact on number concentration and sulfate content of cloud-processed aerosol particles over the North Atlantic. J. Geophys. Res. 100, D10, 21057-21067
INTERLABORATORY COMPARISONS OF THE RESULTS OF ATMOSPHERIC AEROSOLS ELEMENTAL ANALYSIS
O.V.Shuvayeva,K.P.Koutzenogii, V.B.Baryshev, V.I.Rezchikov, A.I.Smirnova, L.D.Ivanova, A.A.Sukchorukov Institute of Inorganic Chemistry; Institute of Chemical Kinetics and Combustion; Budker Institute of Nuclear Physics; United Institute of Geology, Geophysics and Mineralogy, Siberian Branch of Russian Academy of Sciences;
Institute of Nuclear
Physics, Tomsk Polytechnical University, Russia The present work was carried out in frames of "Aerosols of Siberia" project [1], which main goal consists in investigation of formation, transformation and transport of aerosols in Siberia region at the local, regional and global level. Analytical provision of aerosol's investigation composition of aerosols
is one of the important block of the project. An elemental has been established using a
combination of correlation
analysis and statistical data processing to the information produced by different analytical methods. Applied Analytical Methods
Instrumental Neutron Activation Analysis (INAA) Two independent INAA methods have been used: INAA~ and INAA2
Synchrotron Radiation X-ray Fluorescence Analysis (SRXF) Atomic Absorption Spectrometry (AAS) Direct Current Plasma -arc Atomic Emission Spectroscopy (AES) Sampling Aerosols samples were taken on the South of the Novosibirsk region near Karasuk station during the period of February, 5 till March, 3, 1992. The sampling device was an high-volume air intake,which lets a definite air volume pass
through the paper filter
Whatman 41 at a constant pressure without size separation. The results obtained by four independent methods simultaneously( for Cr, Mn, Fe, Ni, Zn, Cu, Ag, Ca, V) are of a greate importance and taken into consideration for the "true" values estimation in present contribution.
722
Comparisons of the results of atmospheric aerosols elemental analysis for P - 0,95 and f -
723
n - 1
As may be seen from the TableI the results for Mn are in line with distribution as o p p o s e d to
Ca-results. Before the rejection of
log-normal
outlyers a correlation
analysis has been applied to experimental results. For each element
the correlation
matrix has been calculated using Basic Statistics & Tables program. The analytical data have been classyfied corresponding to their pair-correlation coefficients K presented in Table II. Table II Pair-correlation coefficients for the elements INAA 1 INAA 1 INAA2
AAS AES SRXF
K>0,33 Mn,Cu Fe,V,A1 Cr, Na Cr,Cu, Fe,AI,V Mn,Na
K0,33 K0,33 K0,33 K0,33 K0,33 the results may consider connected according to linear equation Y - aX + B, where: A and B - coefficients
calculated using the least squares method
taking into account the correlation coefficients. The correlative equation was applied for the data derivatization
to "corrected" results
and consideration them further as
belonging to one and the same general set of analytical data when K>0,33. The results were rejected when K~Z2 t a b . ( P , f )
, two
c o m p a r e d using Barlett e q u a t i o n (~Z2-criterium). W h e n ~Z2 cal. method's
AES
m e a n s in pairs were c o m p a r e d using t-criterium. T h e d a t a
may consider
b e l o n g i n g to the s a m e d i s t r i b u t i o n , w h e n Z2 cal. >Z2 t a b . ( P , f ) a n d teal. < hab.(P, f)
Comparisons of the results of atmospheric aerosols elemental analysis
725
the position of overall mean with it's confidence interval and method means for Ca, including "corrected"result of SRXF, is shown in Fig.2. There is a problem to make a quantitative estimation of the systematic error for each analytical method have been used in present experiment due to the absence of standard samples for atmospheric aerosols, but the developed procedure may be successfully applied for this purpos.
Common
30,000 cm -3) which are considered to have resulted from local photochemical gas-to-particle conversion processes are presented. Factors controlling these production events are examined.
Keywords- background aerosol, gas to particle conversion, aerosol carbon
1 INTRODUCTION, AEROSOL MEASUREMENTS AT MACE HEAD The Mace Head Atmospheric Research Station, situated on the west coast of Ireland (53 ~ N 10 ~ W), has unobstructed access to the Atlantic ocean via a nominal "clean" sector between 180 ~ and 300 ~ Local meteorological data indicate that the majority of air masses arrive at the site via this sector. Aerosol and trace gas measurements from the associated air masses.are considered to be representative of background mid-latitude North Atlantic levels. AN measurements have been carried out at the site using an automatic Nolan-Pollak (O'Connor and McGovern 1991). These measurements have been ongoing since 1989. EBC measurements have carried out using a Aethalometer since 1990 (Jennings et al 1993). Results from measurements carried out during 1991 are presented here. 2 RESULTS OF AN AND EBC MEASUREMENTS Concentration frequency distributions of AN and EBC for 1991 are shown in Figure. 1 and Figure 2 respectively. The AN distribution has a concentration mode at between 100 and 600 cm 3. This mode is considered to be representative of the background AN concentrations at Mace Head. A second smaller mode in the range from 1000-8000 cm -3 is considered to be characteristic of concentrations measured during polluted conditions which occur during periods dominated by continental air masses. Very high AN concentrations, greater than 30,000 cm -3' are also recorded on occasions. These are attributed to local AN production and are discussed in the following section. A distinct bimodal structure is evident in the EBC frequency distribution. The principal mode is considered to be representative of background concentrations and indicates that background values vary between 20 and 80 ng m -3. In a more detailed analysis of these data a seasonal cycle was apparent in the EBC values. However, no seasonal cycle was apparent in the AN concentrations. The seasonal variation observed in the EBC data has been discussed by Cooke (1996). Further results are required before seasonal variation in the AN concentration can be fully assessed. Inter-annual variation generally lead to variations in the relative sizes of the background and polluted modes which are determined by changes in the prevalent meteorology. However, some shifting of the modal mean and median values are also apparent. This is associated with more complex variations in regional meteorology (McGovern, 1996)
734
Aerosol measurements and evidence
of gas-to-particle conversion processes
735
14
25 20
i,~
8
15
6
/
4 ../ a~ 0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
100 Number
10OO concentration
;/
\ ,,
"
~
(?.\
~ s
~,~,
IE4
/
1E5
( n c m 3).3
10
I00
1000
E B C m a s s concentration (ng m "=)
Figure 2 Frequency distribution of EBC concentration for 1991
Figure 1 Frequency distribution of AN for 1991.
3 GAS-TO-PARTICLE PRODUCTION OF AN Occasionally very high AN concentrations are measured at Mace Head. These may occur on a diurnal basis for periods of 3-4 days. Typically the high AN concentrations result from a rapid AN increase following dawn and occur in the absence of similar increases in other measured aerosol species. A sequence of production events from May 1991 is shown in Figure 3 (A). These events are also shown as the rate of Change of the AN concentration in Figure 3(B). The EBC concentration remains relatively low at typical background levels during most of this period indicating that the AN increase is not associated with local anthropogenic pollution. It is of interest to note that the EBC level decreased, on the 14th, as the AN concentration increased.
!
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . .
001,o . . . .
160
~'
140
t
120 100 1000
80 6o 40
-
- - E B C
o
20 100 13/05/91
14/05/91
15/05/91
16/05/91
17105/91
0 18105191
Day
4000 ~ ~
...............................................................................................................
3000
200 180 160
2000
14o
1000
~
120
-1000
13/05/91
- - E B C
80
60 40
-2000 -3000
AN
lOO
o
g o
20 14105191
15105191
16/05/91
17105191
0
18/05/91
Day
Figure 3 AN and EBC values recorded during May 1991 also shown as rate of AN change and EBC. The high rate of AN increase, >3000 N per minute, and the high concentration measured, precludes transport of the AN from other areas or mixed down from a layer aloft as suggested by Ito and Iwai
McGovern et al.
736
(1981) to explain similar measurements in the Southern Hemisphere. The importance of photochemical reactions to AN production is shown from March and April data shown in Figure 4 and Figure 5 respectively. The March data indicates modulation of the daily AN peak concentration associated with variations in the solar radiation intensity. The April 1991 data show that the AN increase is closely linked to sunrise. The AN levels increase rapidly soon after dawn with the peak concentration occurring before midday. Other aerosol measurements during the April 1991, which have been discussed elsewhere (McGovern et al 1996), indicate that the EBC and accumulation mode aerosol (0.1 to 1 pm) remained stable during the first AN peak on the 13th . Subsequent increases in the accumulation mode aerosol, due to local pollution, did not correlate with increases in the AN concentration. However, the AN peaks were reduced by comparison to those recorded on the 13th.
._.
70000 [
...........
I' ...................... I- 700
~ ~176176176176Ii
z
I
50000
11 i
40000 "~
'a 30000~
'~'
~ 15
16
17
:t.
18
19
Day
Figure 3 AN and solar radiation data from March 1991.
:,
i
..
^ ~
I
~E "
AN
1.300 ~ ........ Solar
I 1[.~oo ~
,4t~i !11!
z
"
500 ~_
t 400 o ~
li
~ - ,00000i .................. ~ .................................. ~ 0 0
lOOO
o
z
100! ' 12
!
i t_,, 13
I
~, ! 14 Day
I
i
~,, ; 15
400 ~ w S o l a r
300
~oo~o
~I 0 16
Figure 4 AN and solar radiation data from April 1991.
These data indicate that the high AN concentrations occur as a result of photochemical gas-to-particle conversion processes. The majority of these events have been recorded during the Spring and Summer months but are not confined to these months. AN production events are most prominent when maritime air masses pass over remote land areas. Some of the AN production events occurred in conditions similar to those described by Shaw (1989) which involved the transport of maritime air masses to an inland clean air site. However, it is unlikely that the observations at Mace Head are related to homogeneous nucleation of H2SO4 as suggested in the Shaw study. Production events also occur during polluted conditions. Studies cited by Hoppel et al (1990) indicate that nucleation thresholds of precursor sulphur gases are reduced when NH3 is present. This may be factor in the measurements presented here. 3 CONCLUSIONS Some general characteristics of AN and EBC measurements during background and polluted conditions have been shown and estimate of background concentrations given. The characteristics of AN production events have also been considered. Very high AN concentrations are attributed to photochemical gas-to-particle conversion processes. It is likely that these events were due to nucleation following oxidation of biogenic precursor gases. The exact chemical nature of these precursor gases is not clear but they are thought likely to be sulphur species probably including DMS. Reactive species may also have increased the nucleation potential of precursor gases. The observed decrease in the EBC concentration during AN production may indicate that the absorption characteristics of the ambient aerosol are being modified by coagulative deposition of the AN. Such processes may also affect the hydroscopic properties of the larger aerosol. If such events are typical of coastal boundary areas then they may significantly alter the number concentrations and physcial and chemical characteristics of air masses entering and leaving marine/continental areas.
Aerosol measurements and evidence of gas-to-particle conversion processes
737
REFERENCES Cooke, W.F. (1993): Seasonal variations of the atmospheric absorption cross-section and the relationship between fine mass to elemental carbon,. Atrnos. Environ ,(submitted). Hoppel W.A.,Fitzgerald J.W.,Frick G.M.,Larson R.E. and Mack E.J.,(1990) Aerosol size distributions and optical properties found in the marine boundary layer over the Atlantic Ocean, J. Geophys. Res., 95, 3659-3686. Ito T. and Iwai K., (1981) On the sudden increase in the concentration of Aitken Particles in the Antarctic atmosphere, J. Meteor. Soc. Japan, 59,2,262-271. Jennings S.G., McGovem, F.M. and Cooke, W.F., (1993): Carbon mass concentrations at Mace Head on the West coast of Ireland, Atrnos. Environ., 27A, 1229-1239. McGovern F.M., Jennings S.G,. and O'Connor T.C. (1996) and Simmonds P.G., Aerosol and trace gas measurements during the Mace Head,Atmos Environ, (in print) O'Connor T.C. and McGovern F.M.,(1991 a),Aerosol climatology measurements with a Nolan-Pollak counter, Atmos. Environ. 25A, 561-567. Shaw G. E. (1989) Production of Condensation Nuclei in clean air by nucleation of H2SO4, Atrnos. Environ., 23, 2831-2846.
INFLUENCE OF WATER VAPOR AMOUNT ON SAHELIAN AEROSOLTS SIZE MEASURED AT B AMAKO.
A. KONARE AND C. IROPLO Laboratoire de Physique de 1' Atmosph6re, Universit6 Nationale, Abidjan, C6te d'Ivoire
J. J. BERTRAND Universit6 de Paris XII, France
ABSTRACT The issue of the growth of atmospheric turbidity observed during the rainy season at Bidi(lat=13.916~ ") can be partly answered when studying the relationship between aerosol size distribution and atmospheric water vapor content. The simultaneous measurements of aerosols optical thikness and atmospheric water vapor amount that we carried out with a sunphotometer at Bamako in November and December 1991, provided us with the opportunity to better understand this problem. We showed that the ,i,ngstr6m coefficien ot and I] were strongly related to the atmospheric water vapor content with the respective correlation coefficients r=0.83 and r=0.82. The most likely explanation of the increases of atmospheric turbidity during the rainy season may be the deliquescence growth of the sahelian aerosolts. 1: INTRODUCTION From november, the 5 th to december, the 21st 1991, an experiment of comparison of two kind of sunphotometers was carried out at the Centre R~gional d'Energie Solaire (CRES) of Bamako in Mali. The main goal of that experiment was to investigate some practical and technical characteristics in the use of each kind of sunphotometer as well as to compare them (Konar~, 1992). We used the dataset collected with the two CIMEL sunphotometers. The possibility we had to get from these dataset both aerosol optical thickness and atmospheric water vapor content enabled us to study the influence of water vapor on sahelian aerosols. Therefore, we were able to give a response to the question of the growth of atmospheric turbidity observed during the rainy season at Bidi (picture 1)(Konar~,1995) and at Niamey (Ben Mohamed, 1988). 2: MEASUREMENTS Simultaneous determination of aerosol optical thickness and atmospheric water vapor amounts were carried out. For this purpose, two sunphotometers with I~ each, were used to achieve measurements of solar direct irradiance under cloudless conditions. The first sunphotometer (5F) was provided with five filters of half-widths about 0.011.tm at the wavelengths 0.368 l.tm, 0.44 l.tm, 0.5 gm, 0778 gm and 0.87 gm. The second (6F) was equiped with four filters of 0,011.tm of half-widths for aerosol measurements at 0.44 tam, 0.68 gm, 0.87 l.tm and 1.02 gm and with two filters centred at 0.94 gm with different half-widths:0.01 l.tm and 0.05 gm for water vapor measuremens. The sunphotometers were calibrated by the use of the modified Langleyplot method (V Soufflet et al, 1992).
738
Influence of water vapor amount on Sahelian aerosols size measured at Bamako
739
2-1: AEROSOLS OPTICAL PARAMETERS DETERMINATION Computations of atmospheric turbidity were made according to the usual Bouguer-Lambert-Beerlaw expressing the measured intensity at wavelength ~,. Ix = Iox " e-m~)k, as a function of the calibration constant Iox, the compared airmass m and the total atmospheric optical depth "cx. From a:x, the optical thickness of aerosols particles is deduced "tax = "k -q:Rx -'l~03X "~Rx and a:O3x are respectively the Rayleigh optical thickness due to the scattering by air molecules and ozon optical thickness were calculated using 5S code (Tanre et al, 1988). The well-known ,~nstr6m formula: "tax = 13~"~ gives the ,~ngstr6ms exponent coefficient ct and the Anstr6ms turbidity coefficient ]3. 2-2 WATER VAPOUR AMOUNT DETERMINATION We used the differential absorption technique of Frouin et al (1990) to deduce the total atmospheric water vapour amounts. The two channels employed were centred at 0.94 gm with 0.01 and 0.05 gm bands. The ratio of the transmission functions, for the narrow and the wide band channels was computed with the lowtran 7 code (Kneizys et al, 1988) for different airmass and using the tropical model of Mac Clatchey et al (1971). We deduced the atmospheric water vapour amount from an interpolation method (Konare, 1995). 3: RESULTS AND DISCUSSION The important aerosol optical thickness observed at Bidi during the rainy season in picture 1, presume an important transfer of aerosol upon the moonsoon layer. During the period of measurements we obtained a large variation of the Angstr6ms coefficients respectively from 0.3 to 0.9 for the ,Xmgstr6m exponent ot and from 0.3 to 0.9 for the Angstr6m turbidity coefficient 13, while the atmospheric water vapour amount varied from 1.45 to 2.8 (g/cm2). In picture 2, we show 168 measurements that both the aerosol size caracterized by o~ and the atmospheric turbidity caracterized by 13 increase with the atmospheric water vapour amount according to the relationship: o~ = -0.779 u + 2.254 and 13 = 0.681 u - 0.833 with the correlation coefficients r = 0.83 and r = 0.82 respectively. In an earlier study, Bertrand (1977) emphazised a fair hygroscopic character of saharan aerosols. The hygroscopic aspect of sahelian aerosols that we point out in this study, could be explained by the presence of a new aerosol type in relation to new sources of aerosols in the sahel region, especially in the upper region of the niger river (N'tchayi et al, 1994), (Iroplo, 1995). 4: CONCLUSIONS The main goal of this paper has been to point out an hygroscopic aspect of sahelian aerosols. The results presented in this paper confirm the capability of sahelian aerosols to be good condensation nuclei which allows them to play significant role in the precipitation process via a comprehensive approach of the complex interaction aerosol-cloud. This study is also a contribution in the monitoring of the climatic impact of the sahelians" aerosol-induced cloud which remains poorly
740
Konare et al.
understood. The direct consequence of the increases in diameter due to deliquescence growth of sahelian aerosols is the modification of their refractive index. This effect needs additional up to date studies on the internal structure and the shape of the sahelian aerosols which results will be included into climate models. These studies will lead to a meaningfull interpretation of the changes in the sahelians" aerosols growth behavior with respect to the appearance of new aerosol sources. 5: REFERNCES Ben M., A., (1988) Contribution ~ l'Etude de l'aErosol sahElien au Niger. ThEse de Doctorat d'Etat, UniversitE de Niamey, Niger. Bertrand, J. J., (1977) Action des poussiEres subsahariennes sur le pouvoir glacogEne de l'air en Afrique de l'Ouest. ThEse d'Etant n ~ 253, Universit6 de Clermont Ferrand. Frouin, R., P.Y. Deschamps and P. Lecomte, (1990) Determination from space of atmospheric total water vapour amounts by differntial absorpyion near 940 nm: Theory and airborne veification. J.G.R., 95 13927-13935. Iroplo, C., (1995) Essai de caracterisation des zones potentielles de deflation. ThEse de troisiEme cycle, n ~ 220, UniversitE Nationale de C6te d'Ivoire. Kneizys, F.X.E.P. Shettle, W. O. Gallery, J.H. Chetwind, Jr., L.W. Abreu, J.E.A. Shelby, R.W. Fenn and R.A. Mc Clatchey, (1988) Atmospheric transmittance/radiance: computer code Lowtran-7. AFGL. Environ. Res. Pap., N ~ 697, AFGL-TR-80-0067. KonarE, A., (1992) Rapport d'intercomparaison des photomatres Cimel et Noll. Konard, A., (1995) Contribution ~ l'implantation d'un rEseau de photomEtres en Afrique: Etude critique et exploitation des mesures: cas de Bidi et Bamako., these de troisiEme cycle, n~ I, Universitd Nationale de CEte d'Ivoire. McClatchey, R. A., R. W. Fenn, J.E.A. Selby, J. S. Garing et F. E. Voltz, (1971) Optical properties of the atmosphere. (AFCRL-71-0279) Air Force Cambrige Research Laboratories, Bedford Mas. N'Tchayi, M.G., J.J. Bertrand, M. Legrand, J. Baudet, (1994) Temporal and spatial variations of atmospheric dust loading throughout West Africa over last thirty years. Anneles Geophysicae., 12, 265-273. Souffiet V., C. Devaux et D. Tanr~, (1992) Modified Langley plot method for measuring the spectral aerosol optical thickness and ots daily variations. App. Opt 31,12. TanrE D., Deroo C., Herman M., Mocrette J., Perbos J., Deschamps P.Y., (1990) Description of a computer code to stimulate signal in solar spectrum. Int. J.R.S., vol 11, pp 659-668.
Influence of water vapor amount on Sahelian aerosols size measured at Bamako
[]CX
ol3 1
1
+
I
i
+
0J
J I
0
J
A
J
,
0
J
A
J
==...v
Months
87
88
Fig 1 Time evolution of the ,~ngstr6m coefficients oc and [3. The months inside the frame correspond to the rainy season. o (l = -0.779 u + 2.2_54 (r=0.83) 000o 8 o
.9
o
o o r
~
o
.8
o g
o
o
o ~ = 0.681 u- 0.833 (r=-0.82)
o~o
o o
o ~ o
~ oo o o ~176 ~
o
o
o
o o o ~ o o--......2o ~ . ~
o
~
o
.7
oo
o
.~
+6
-~~
8
~
o
: ~ o- -_o~o ~~176
o
8o
.4
9
.2
1.6
+
nn~
.
1.7
-~
1.'8
1;9 2.O 2;1 water vapor amount (g/m2) Fig 2: ,~ngstr6m coefficients r and 13as functions of the atmospheric water vapour amount u.
2;2
741
ACE-l: A FIELD STUDY OF ATMOSPHERIC CHEMISTRY
AEROSOL PHYSICS AND
J.L. GRAS CSIRO Division of Atmospheric Research, PMB 1, Aspendale, Vic., 3195 Australia T.S. BATES NOAA-PMEL, 7600 Sand Point Way NE, Seattle, WA 98115 USA B. J. HUEBERT Department of Oceanography, University of Hawaii, Honolulu, HI 96822, USA
Abstract - ACE-1 was a field study developed within IGAC in response to the problem of reducing the current uncertainty in calculating climate forcing by atmospheric particles. Particular emphasis was placed on detailed characterization of clean marine boundary layer aerosol, processes related to production and evolution of this aerosol and measurements of the radiative effects of the aerosol. The experimental phase of the study, which was centred on the Southern Ocean south of Australia, was successfully completed between Nov. 15 - Dec. 15 1995. It incorporated coordinated measurements from the US NCAR C-130 aircraft, the NOAA research vessel Discoverer, the Australian fisheries research vessel Southern Surveyor, Cape Grim and Macquarie Island, and involved over 100 scientists from 11 countries. Keywords- ACE-1; climate forcing; remote marine boundary layer; aerosol characterization INTRODUCTION Accurate assessment of the climatic effects of atmospheric aerosol particles is essential for policymakers to plan responses for possible natural and human-induced changes in the chemistry of the atmosphere and global climate. Whilst the past couple of decades have seen a rapid expansion in our ability to generate predictive climate models, these efforts have not necessarily been matched by a corresponding ability to adequately describe our physical and chemical environment. In particular, there is a marked lack of knowledge of the global distribution of aerosol properties and the processes controlling these properties. This lack of knowledge severely limits our ability to reduce the uncertainty of modelled aerosol climate-forcing to an acceptable level. The basic processes whereby aerosol particles affect climate are relatively straightforward and well known in principle; direct effects are the result of scattering and absorption of solar radiation by particles, indirect effects are primarily those due to modification of cloud optical properties through aerosol-dominated cloud-droplet nucleation. Based on current published model estimates for aerosol effects on climate (IPCC 1994) the direct effect of anthropogenic sulfate aerosol is believed to be
742
ACE-l: A field study of atmospheric aerosol physics and chemistry
743
responsible for a climate forcing the range -0.25 to -0.9 W m -2 and aerosol from biomass burning for a forcing in the range -0.05 to - 0.6 W m- . Indirect effects through cloud modification are less certain, falling within a range of 0 to - 1.5 W m -2. The sum of these components allows a very significant negative forcing although it is clearly essential that the uncertainty is reduced. Aerosol particles are intrinsically difficult to study and to categorise. They have a multitude of sources including natural processes and nearly all human activities. They exhibit a continuous spectrum of sizes from nanometre to tens or even hundreds of micrometre and their composition varies with size; their radiative importance and cloud nucleating ability also depends critically on size as well as composition. As a further complication, there is continuous evolution of sizes as particles respond to and modify their environment. Collation of the necessary data on a global scale clearly needs a major international effort.
IGAC ACAPS - AEROSOL CHARACTERIZATION AND PROCESS STUDY ACTIVITY One of the clearest early milestones in recognition of the potential role of aerosol in climate was the SCEP and SMIC studies in the early 1970s (Study of Critical Environmental Problems and Study of Man's Impact on Climate, MIT 1970,1971), presaging current concern about both direct and indirect climate effects of particles. These panels also pointed to the need for directed research and extensive measurement programs. Since that time considerable advances have been made both in measurement and modelling although the role of coordinated international programs has been relatively limited until quite recently. The International Global Atmospheric Chemistry Program (IGAC) was established at a meeting in Dookie Australia in 1988. From the outset IGAC has contained a number of aerosol related activities with early interests directed mainly at cloud condensation nuclei (CCN) and marine aerosol and gas exchange (MAGE). Several evolutionary steps have followed - in 1991 the CCN activity was revised to include a much broader range of aerosol related issues becoming the Multiphase Atmospheric Chemistry (MAC) activity. A major advance occurred in 1995 when the Intemational Global Aerosol Program (IGAP), which was formalised in 1994 merged with IGAC. One outcome of the merger is a much higher profile for aerosol research within IGAC through an expanded "Focus on Atmospheric Aerosol". This new focus comprises four areas - three of which were previously incorporated to some extent in the IGAC MAC activity and a totally new activity associated with stratospheric aerosol. These new (tropospheric) activities have been set up to foster research into direct aerosol radiative forcing (DARF), aerosol-cloud interactions and (their role in) climate (ACIC) and aerosol characterization and process studies (ACAPS). Several other activities within IGAC, including the global modelling activity (GIM), have also been augmented. Subsequent discussion here will be largely directed to the ACAPS activity and field programs developed through this activity and its predecessor IGAC MAC. As initially established there are three principal goals for IGAC ACAPS, these are: 1/understanding the spatial and temporal distributions of aerosol properties and their interrelationships 2/determination of the processes controlling the formation and fate of aerosol and how these processes affect the aerosol (size, number, composition - and dependent properties) 3/examination of the role that aerosol has on other atmospheric species and better understand the interaction of atmospheric aerosol with the major biogeochemical cycles (S, N, C, trace metals)
744
Gras et al.
Achieving these goals can be broken into broad two tasks - the first aims at improving aerosol characterization the second at improving the existing understanding of processes involving aerosol production, evolution and removal. In order to achieve any real progress within these broad areas it is necessary to focus on a smaller set of key objectives. Considering characterization initially, the first objective is to develop and use closure studies to examine internal consistency of different measurement and modeling strategies. Closure studies represent a deliberately overdetermined set of measurements with specific objectives of ascertaining the level of uncertainty in measuring or modelling a particular aerosol property, calibration of remote sensing techniques or identifying where improvements in measurement or modelling are required. The second objective, which is not totally independent from the first, is to develop standardized sampling and analysis procedures - this is seen as essential for obtaining high quality, comparable data sets. The emphasis here is equally on comparable as high-quality, since even apparently minor differences in sampling strategy, for example whether size discriminating aerosol collection stages are operated at ambient or under dry conditions, can make relatively large differences in the mass collected. There are many other possible examples. The main means advocated and developed by the contributing science teams in IGAC ACAPS and formerly MAC for implementing these objectives is a series of field programs that have the acronym ACE, for Aerosol Characterization Experiments. These experiments are intensive campaigns in regions that will hopefully represent the main atmospheric regimes, encompassing clean maritime conditions through continentally impacted maritime to fully continental. The second broad task within ACAPS, that of working towards improved understanding of aerosol production, evolution and removal processes also requires focussed effort for advancement. Areas seen as potentially benefiting from an international program approach include a number of specific chemical/physical processes related to aerosol formation and its subsequent evolution. An additional component of this work is the active promotion of the development of process models describing gasparticle conversion and aerosol evolution. Implementation of this task also centres primarily around the ACE field experiments as venues for the detailed field measurements that are necessary to advance the process models.
ACE EXPERIMENTS The first of a series of these studies, ACE-1 took place from Nov. 15 - Dec. 15 1995 in minimallypolluted conditions, focussing particularly on the marine boundary layer over the Southern Ocean south of Australia. This experiment coordinated research contributions of over 100 scientists from 11 countries, making measurements from the US NCAR C-130 aircraft, the NOAA research vessel Discoverer, the Australian fisheries research vessel Southern Surveyor, and land based stations at Cape Grim (40.7 ~ 144.7 ~ and Macquarie Island (54.5 ~ 159.0 ~ The science team identified three main objectives for ACE-1: 1/Documentation of.the chemical, physical and radiative characteristics of remote marine aerosol and investigation of the relationships between these aerosol properties. 2/Determination of the key physical and chemical processes controlling the formation and fate of aerosols and how these processes affect the number, size distribution, chemical composition and the radiative and cloud nucleating properties of the particles.
ACE-l: A field study of atmospheric aerosol physics and chemistry
745
3/An assessment of the climatic importance of remote marine aerosols. Within these objectives a series of more specific scientific questions or issues were proposed for the ACE-1 experiment to address. The first of these issues is establishing the precision with which the radiative and cloud nucleating properties of an aerosol can be derived from its measured physical and chemical properties. This entailed a series of seven types of intercomparison or local closure experiments coveting aerosol mass, size - humidity dependence, scattering coefficient and its humidity dependence, refractive index, CCN activity, and CCN - cloud drop number relationship. Other basic measurements included aerosol size distribution and size-segregated chemical composition. A further over-determination involved duplication of as many as possible of these measurements on all land, ship and airborne platforms. Extending the closure principle to a vertical column in the atmosphere, to give a column closure experiment is necessary to address the second issue. This relates to establishing the level of accuracy with which measured physical and chemical properties of the aerosol, taken over a vertical column can be used to predict the integrated effect of aerosols on radiative transfer. For clear sky cases, measurements include aerosol microphysical properties, upward and downward radiances at altitude, aerosol optical depth, lidar profiles of backscatter from the aircraft and satellite (AVHRR) radiances. ACE-1 allowed low optical depth case studies over the Southern Ocean, where optical depths are typically around 0.04 -0.05 (Forgan 1990, Durkee et al. 1991) and by incorporating an additional study near Hawaii during the C-130 aircraft transect from the USA to southern Australia, a high optical depth case in a relatively well-defined volcanic plume from Kiluea was also possible. Prediction of aerosol concentrations and size distributions from the precursor level requires understanding of a series of cascading processes. Three of these were identified as particular issues for attention in ACE-1. The first seeks an understanding of the biological, chemical and physical processes controlling the concentration of dimethylsulfide (DMS) in surface seawater and its flux to the atmosphere. This is a good example of the multi-disciplinary approach that is ultimately required for predicting atmospheric aerosol properties, requiting knowledge of sea-water sulfur species concentrations, details of the phytoplankton species, physical oceanographic parameters, meteorology and atmospheric DMS and photochemically active species concentrations. The second process study targeted for ACE-1 is the chemical evolution of aerosol particles in the marine boundary layer; the experiment aimed to establish the rates and efficiencies of the processes controlling DMS and SO 2 oxidation. This is a potentially very-difficult experiment; in order to separate dynamical and chemical processes, measurements need to be taken for an extended period in a Lagrangian framework, following a known air parcel. Because this experiment was aircraft based, high-speed high-precision measurements were required for gas phase DMS oxidation products, SO 2, methanesulfonate (MSA) and H2SO 4 as well as concentrations of photochemically active species, aerosol chemistry and aerosol particle size distributions over the diameter range of 3 nm to 10 ktrn. The Lagrangian study also required knowledge of sea-water DMS concentrations along the experiment track, which were obtained using the two research ships, Discover and Southern Surveyor. Following identifiable air parcels over the Southern Ocean presents a severe challenge and was achieved by using a smart balloon package (Businger et al. 1996). Balloons with the smart package were released from the NOAA ship at the start of each experiment to follow the selected an air parcel; once operating they broadcast their location as derived by GPS navigation system allowing remote tracking and relocation from the C-130 aircraft to enable multiple sampling of the specified air parcel over several days.
746
Gras et al.
Whereas the previous two process studies relate directly to the chemical evolution of marine boundary layer aerosol the third class of process study was concerned with the rates and efficiencies of processes controlling the nucleation, evolution of particle size-distributions. A still unresolved question for the marine boundary layer aerosol is the origin and conditions surrounding the production of new particles that maintain a relatively constant background concentration, seen for example at remote locations such as Cape Grim (Gras 1990). Usually a large existing aerosol surface area inhibits new particle production although for some specific conditions indications of new particle production have been reported (Ito 1985, Covert et al. 1992, Gras 1993). The free troposphere was suggested as a possible reservoir regulating marine boundary layer (MBL) particle concentrations by Bigg et al. (1984); recently model calculations by Raes et al. (1993, 1995) have identified the upper troposphere as a potential region for new particle production. In-situ measurements in the upper troposphere by Clarke (1993) provided direct evidence of highly-abundant ultra-fine (3 nm< D < 20 nm) particles adding further support to this hypothesis. Marine clouds have previously been identified as regions of new particle production (Hegg et al. 1990, Perry and Hobbs 1994). Clouds are also important in the general evolution of particle size distributions in the MBL (Hoppel et al. 1986). As air parcels cycle through non-precipitating cloud, activated aerosol becomes subjected to rapid aqueous phase oxidation processes allowing the development of a distinct and characteristic size mode on cloud droplet evaporation. Measurements of the effects of cloud cycling on the size distribution afford a means for establishing the fraction of sulfur species oxidised in cloud. State of the art systems for detecting and sizing ultra-fine particles and measuring trace levels of gaseous precursors including H2SO 4 MSA and NH 3, along with a wide range of other supporting aerosol chemical and physical measurements were included in the ACE-1 aircraft instrument complement. During ACE-1 these systems were used very successfully to investigate the conditions of new particle production, perhaps most spectacularly in the free troposphere in cumulus outflows. New particle production was also observed over extended regions in the MBL from observations at Cape Grim, Macquarie Island, and over the Southern Ocean from the NOAA ship Discoverer; all of these platforms were instrumented for aerosol microphysical and chemical measurements. Production was found to be associated with mixing events following frontal passages and during periods of synopticscale subsidence and convective mixing between the free troposphere and the marine boundary layer. The final broad objective of ACE-1 has one specific question or issue, how the observations from ACE-1 can be used to improve the accuracy of aerosol climate models. An important aspect of the planning of ACE-1 has been close interaction with members of the modelling community to develop an experiment that provides data that are indeed useful for validating and refining models. We anticipate that data from ACE-1 will be used in a hierarchy of models starting with detailed process models of DMS production and exchange into the atmosphere, atmospheric chemical reaction rates and aerosol dynamics models used for predicting size evolution. ACE-1 data should also find application in one dimensional models that need vertical prof'lles of aerosol properties for radiative transfer calculations. The combination of extensive cloud microphysical and aerosol measurements obtained from the C130 should also be of particular use for testing three dimensional large-eddy simulation models that assess the interaction between aerosol and clouds. Finally, ACE-1 data from an otherwise data sparse region over the Southern Ocean will find application in verification of model outputs from regional and threedimensional global models that predict geographic and temporal distributions of aerosol species.
ACE-l: A field study of atmospheric aerosol physics and chemistry
747
CONCLUSIONS During ACE-1 an intensive series of chemical, physical, radiative and cloud nucleating properties of aerosols were made from surface, ship and airborne platforms in a wide range of environments. This included the background marine atmosphere, volcanic plumes of Kilauea Hawaii and Mt. Ruapehu New Zealand, biomass burning and anthropogenic air masses from Australia. Transits of both the NCAR C130 and the NOAA ship Discoverer afforded a unique opportunity to extend the range of detailed aerosol and related measurements from 75 ~ to 60 ~ over the Pacific and Southern Oceans for altitudes between 0 - 6 km. The resulting data now form the basis for closure studies that assess our ability to calculate radiative and cloud nucleating properties of the aerosol from its fundamental properties, the chemical and physical size distributions. Biological and chemical characteristics of the water masses south and west of Tasmania and their evolution with time were studied during the course of the experiment. Initial results suggest biological productivity and DMS production were highest where seasonal thermoclines were being established and where mesoscale mixing of waters of subtropical and subantarctic origin was occurring. These data will also be used to assess processes controlling the concentrations of DMS and the rates of release and transfer of DMS to the atmosphere. Despite the obvious problems of conducting a Lagrangian experiment in the "roaring forties" where wind and weather conditions are known for their severity, two successful Lagrangian missions were carried out, each over two days. As with previous ASTEX/MAGE studies these will be used to quantify the rates and efficiencies of sulfur gas oxidation and the evolution of the aerosol. Although the ultimate success of the ACE-1 experiment can only be fully gauged after a more complete data analysis period, it is very clear from preliminary analyses that most of the experiment's initial objectives have been met or exceeded.
REFERENCES Bigg E.K., J.L. Gras and C. Evans (1984) Origin of Aitken particles in remote regions of the Southern Hemisphere. J. Atmos. Chem., 1,203-214. Businger S., S. Chiswell and W.C. Ulmer (1996) Balloons as a measurement platform for atmospheric research. J. Geophys. Res., 101, 4363-4376. Clarke A.D. (1993) Atmospheric nuclei in the Pacific mid-troposphere: their nature, concentration and evolution. J. Geophys.,Res., 98, 20,633 - 20,647. Covert D.S., V.N. Kapustin, P.K. Quinn and T.S. Bates (1992) New particle formation in the marine boundary layer. J. Geophys. Res., 97, 20,581-20,589. Durkee, P.A., F. Pfeil, E.M. Frost and R.A. Shema (1991) Global scale aerosol particle characteristics from satellite detected radiance. Atmos. Environ., 25A, 2457-2465.
748
Gras et al.
Forgan B.W. (1990) Aerosol optical depth, In Baseline 88, Eds. S.R. Wilson and G.P. Ayers, Bureau of Meteorology and CSIRO Division of Atmospheric Research, pp 61-62. Gras J.L. (1990) Baseline Atmospheric condensation nuclei at Cape Grim 1977-1987. J. Atmos. Chem., 11, 89-106. Gras J.L. (1993) Condensation nucleus size distribution at Mawson Antarctica: seasonal cycle. Atmos. Environ., 27A, 1417-1425. Hegg D.A., L.F. Radke and P.V. Hobbs (1990) Particle production associated with marine clouds, J. Geophys. Res., 95, 13,917-13,926. Hoppel W.A.., G.M. Frick and R.E. Larson (1986) Effect of non-precipitating clouds on the aerosol size distribution in the marine boundary layer. Geophys. Res. Lett., 13, 125-128. Ito T. (1985) Study of background aerosols in the Antarctic troposphere. J. Atmos. Chem., 3, 69-91. IPCC (1994) Climate Change 1994: Radiative forcing and an evaluation of the IPCC IS92 Emission Scenarios, F_As.J.T. Houghton, L.G. Meira Filho, J.P. Bruce, Hoesung Lee, B. A. Callander, E.F. Haites, N. Harris and K. Maskell. Cambridge University Press, Cambridge, UK. Massechusetts Institute of Technology (1971) Report on the Study of Man's Impact on Climate (SMIC), MIT press, Cambridge. Massechusetts Institute of Technology (1970) Report on the Study of Critical Environmental Problems (SCEP), MIT Press, Cambridge. Perry K.D. and P.V. Hobbs (1994) Further evidence of particle nucleation in clean air adjacent to marine cumulus clouds. J. Geophys. Res., 99, 22,803 -22818. Raes F., R. Van Dingenen, J. Wilson and A. Saltelli (1993) Cloud condensation nuclei from dimethylsulfide in the natural marine boundary layer: remote vs. in-situ production. In: Dimethylsulfide, Ocean, Atmosphere and Climate, Eds. G. Restelli and G. Angeletti, Dordrecht, pp 311-322. Raes F., J. Wilson and R. Van Dingenen. (1995) Aerosol dynamics and its implication for global aerosol climatology. In Aerosol forcing of climate. Eds. R.J. Charlson and J. Heintzenberg. John Wiley & Sons, Chichester.
ASPECTS
OF TROPOSPHERIC QUESTIONS
AEROSOLS: AND ACE-2
FRANK RAES Environment Institute, European Commission Joint Research Centre, I- 21020 Ispra (VA) Italy Abstract - A number of questions related to atmospheric aerosols will be tackled by IGAC's second Aerosol Characterization Experiment (ACE-2, Tenerife, summer 1997). In this paper we present these questions and review them in the light of the latest literature. Direct and indirect radiative forcing by anthropogenic aerosols is the main topic of ACE-2. We stress the importance of a complete aerosol charaterization (chemical mass closure) as a basis for any radiative forcing calculation. We propose a better focus on the effect of mixed aerosols, in particular aspects of mixing anthropogenic aerosols with dust, and an opening towards the role of aerosols in atmospheric chemistry in general.
INTRODUCTION IGAC's second Aerosol Characterization Experiment (ACE-2) will be conducted from 16 June to 25 July over the sub-tropical Northeastern Atlantic Ocean. Its goal is to determine and understand the properties and controlling factors of the aerosol in the anthropogenically modified atmosphere of the North Atlantic and assess their relevance for radiative forcing. Aerosols are currently prescribed entities in General Circulation Models (GCM). Aerosol size distributions and chemical compositions, based on limited observational data, are frozen in a global static aerosol climatology. Mie theory is then used to derive aerosol radiative properties and to calculate the direct radiative effect of various aerosol types (Schult and Koepke, 1994). The radiative effect of clouds is calculated in GCM's in terms of an effective cloud droplet radius, which is often parahaeterized independenty of the aerosol loading ( Martin et al., 1994). Steps in the direction of on-line, dynamic aerosol climatologies have been undertaken: emission, chemical transformation and removal of individual aerosol components (and their precursors) have been described in global transport models and GCM's. This has resulted in global fields of the mass concentration of sulfate (e.g. Langner and Rodhe, 1991, Pham et al., 1995 ), black carbon (Cooke and Wilson, 1996), desert dust (Tegen and Fung, 1995), on which several estimates of the global direct and indirect forcing have been based (see below). ACE-2 aims at a more realistic description of aerosols in GCM's, based on a better understanding of the processes that determine the aerosol number concentration, size distribution, chemical composition and the state of mixing of the atmospheric aerosol. IN-SITU AEROSOL CHARACTERIZATIONS IN ACE-2 Question 1. Can the measured physical and chemical properties of the atmospheric aerosol be used to predict the radiative and cloud nucleating properties of that same aerosol? The central role played by the aerosol distribution with respect to size and chemical composition in calculating properties relevant to radiative forcing is known (e.g. Raes et al., 1995). In principle, it requires multi-dimensional distributions that describe the number of particles not only in terms of their size, but also in in terms of their refractive index (m) or fraction of soluble material (e). Both m and e are related to the size-dependent chemical composition of the aerosol.
749
Raes
750
Obtaining closure between direct measurements of integral properties like CYxt(X,r,t) and CCN(Sc,r,t ), and their values derived from the distributions N(Dp,m,r,t) and N(Dp,~,r,t) is required for reducing the uncertainty in aerosol radiative forcing calculations. So far that has only been possible in conditions where the multi-dimensional distributions N(Dp,m,r,t) and N(Dp,~,r,t) simplify to size distributions N(Dp,r,t) with m or e constant within a given size interval (Quinn et al., 1994). Recent measurements have shown that this will not always be the case (see Heintzenberg and Covert, 1990, for a review). For instance, in aerosols from anthropogenic origin, the fraction of soluble material ~ in particles of a given size is not constant (Str6m et al., 1992, Svenningson et al., 1992). Such characteristics result from how aerosols from different sources mix, and from the condensation of nonvolatile particulate matter on the particles. Because of the lack of observations on the state of mixing in aerosols, mixing is usually treated superficially in radiative transfer calculations. Calculations however show it can have an important effect when it comes to absorption by aerosols (Heintzenberg, 1978; Chylek et al., 1995). A major breakthrough is needed in assessing the role of organics in the uptake of water by aerosols, and eventually their optcial and cloud activation properties. Observations show that they can either increase or diminish the uptake of water, with respect to the amount calculated based on the inorganic species only. (Saxena et al., 1995). Shulman et al. (1996) have extended the Kohler theory and offered a theoretical framework to account for organics in cloud activation.
Question 2. Are there useful empirical correlations between the mass concentration of individual aerosol components and the radiative and cloud nucleating properties of that same aerosol? It is not yet possible to treat (multi-dimensional) aerosol distributions as prognostic variables in global climate models and do an 'ab-initio' calculation of the effect of aerosols. The present estimates of the radiative forcing by anthropogenic aerosols are based on calculated global 3-D fields of the mass concentration of individual aerosol components, using empirical "mass - effect" relationships and/or assumptions regarding the aerosol size distribution (direct forcing: Charlson et al.,1991, Kiehl and Briegleb, 1993, Boucher and Anderson, 1995, Haywood and Shine, 1995, indirect forcing: Jones et al.,1994, Boucher and Lohman,1995) This approach has been important in obtaining the first estimates of radiative forcing by sulfate aerosols. Very likely, similar approaches will be adopted to estimate the radiative effects of other aerosol species. For the case of the direct effect, sensitivity studies have shown that the calculated forcing is only weakly dependent on the particle size distribution and chemical composition (Kiehl and Briegleb, 1993, Boucher and Anderson, 1995, Nemesure et al., 1995, Pilinis et al., 1995) such that a direct forcing based on mass fields might be sufficiently accurate. The range of applicability of these calculations must be tested as a part of ACE-2, using the various aerosol types (clean marine, anthropogenic, desert dust) and the wide range of aerosol concentrations of compositions expected in the study region. A fundamental piece of information is how the major aerosol components contribute to the total aerosol mass (chemical mass closure); little new information has become available after the compilation by Heintzenberg (1989).
Question 3. Can the measured physical and chemical properties of the aerosol in the vertical be used to accurately predict the column-integrated direct effect of aerosols on radiative transfer? Radiation is transmitted through the whole depth of the atmosphere, and changes of the albedo at the top of the atmosphere (TOA) are a result of changes in the column integrated radiative
Aspects of tropospheric aerosols: Questions and ACE-2
751
properties of the atmosphere. Radiative transfer calculations will be tested by comparing vertically integrated LIDAR information with columnar extinction data derived from ground-based and air- and space-borne radiometers, and with in-situ measured physical, chemical and radiative properties of the aerosols in the column. A direct measurement of the flux divergenve accross an aerosol layer (hence, direct radiative forcing) will be attempted. Calculations show that such a closure study is feasible in the continental plume advecting from Europe into the ACE-2 area (Russell et al., 1996). The major aerosol effect on the radiation balance over the tropical Atlantic, and possibly in the ACE-2 area, is caused however by Saharan desert dust (Husar and Stowe, 1995, Li et al, 1996). Saharan dust might have an anthropogenic contribution (Tegen et al., 1996) due to changing agricultural practices and because it might be mixed and modified by pollution from Europe (Prospero, et al., 1995). This will complicate column closure measurements. However mixing of anthropogenic aerosols with dust aerosols might not be an uncommon feature (e.g. Gobi desert dust mixed with anthropogenic aerosol from China) and the reduction of the scattering efficiency of e.g. sulfate mass when it is internally mixed with large dust particles must be studied (Dentener et al., 1996)
Question 4. Can the measured physical and chemical properties of aerosols and clouds in the vertical column be used to accurately predict the integrated indirect effect of aerosols on radiative transfer? The change of the albedo of a cloud is the result of changes in the microphysical properties throughout the cloud depth, which in turn is controlled by the change of the number of aerosol particles that are activated in the cloud (Twomey et al., 1984.). Additional mechanisms seem to be able to reinforce this change of albedo through inhibition of precipitation, thus leading to increased horizontal cloudiness (Albrecht, 1989) as well as increased cloud thickness (Pincus and Baker, 1994). Present estimates of the indirect radiative forcing by aerosols is highly uncertain because of the inadequate knowledge of sources, properties and distribution of aerosols (as for the direct effect), and because of the insufficient understanding of the processes by which aerosols interact with clouds and eventually effect their optical properties. A particular problem is the 'anomalous absorption' of fight in clouds, which is an obstacle to final closure, since it can account for any discrepancy between observed and calculated radiative fluxes. Measured absorption is usually found to be higher than the calculated one (Cess et al., 1995,, Wiscombe, 1995). It has not been established whether this anomalous absorption is real and due to absorption by cloud droplets or interstitial aerosol, or whether it is due to 3-D cloud inhomogeneities. Different estimates of the indirect effect give different importance to the sensitivity of low level clouds topping sub-tropical MBL's. The estimates of Boucher and Lohman (1995) show hardly any effect in such cloud systems, whereas those of Jones et al. (1994) do. This might be a consequence of the way the MBL and low clouds are described in such systems, or a consequence of the way the effect of aerosols on clouds is parameterized. In the case of the ACE-2 area, models also seem to have difficulties to capture transport out of Europe, southwards into the area (Putaud, et al., 1996) AEROSOL PROCESS STUDIES
Question 5. What are the rates and efficiencies of the processes in the cloud free MBL that change the number size distribution and size-dependent chemical composition of a continentally derived aerosol as it advects over the North Atlantic? The processes that shape the number size distribution and size-dependent chemical composition of the aerosol, and hence its radiative impact are qualitatively known (e.g. Hoppel, 1990). They include
752
Raes
homogeneous gas phase chemistry, aerosol dynamics (nucleation, condensation, coagulation, dry deposition), cloud processing (activation, in-cloud chemistry, coalescence, evaporation and precipitation). Depending on the meteorology during the experiment, on process might be dominating the other. During ASTEX, the particle number budget was found to be dominated by entrainment, which caused dilution by MBL growth and divergence in the horizonta flow field (Clarke et al., 1996). On this occasion the European plume was advected within the MBL. However, when the continental aerosol is transported in the free troposphere, as usually is the case with desert dust, entrainment could be a major source of aerosol in the MBL. Aerosol mass and number fluxes across the top of the MBL, and more accurate entrainment rates must be quantified to close the aerosol mass and number balance of the MBL. Presently, entrainment rates have been estimated only within a factor of two (Bretherton,
1995, A yers and Galbally, 1994). Homogeneous nucleation of precursors has been observed in the clean MBL (Covert et. al., 1992) and more in particular in coastal regions (McGovern et al., 1996). However, the occurrence of nucleation is subject to a variety of conditions such as pre-existing aerosol surface area. When the conditions are not fitUfilled, the precursor will condense onto the pre-existing aerosol. The latter has been the case in a number of field experiments of which we mention only those in the subtropical N. Atlantic (Clarke et al., 1996, Raes and Van Dingenen, 1996, Van Dingenen et al., 1996). Theory doesn't adequately deal with aerosol nucleation and growth processes. Predictions of aerosol dynamics models, even when applied to the simple case of the clean MBL, disagree strongly about the importance of in-situ nucleation (Pandis et al., 1994, Raes, 1995, Raes and Van Dingenen, 1995). This leaves a lot of room for speculation: e.g. the role of ammonia in the nucleation process has been roughly estimated (Coffman and Hegg, 1995), as well as the role of co-condensation of HCI and HNO3 vapours in particle growth (Kerminen et al., 1996). Sea spray is an important source of aerosols over the oceans. It has usually been assumed that sea spray produces large (> 1 gm) particles, that have a residence times which are too short to play a role as CCN (Charlson et al, 1987). However, recent observations have re-opened the debate about the role of seasalt and other sea-surface derived particles in the aerosol and CCN number concentration
(O'Dowd et al. 1993). It has been argued that oxidation of SO 2 in sea-salt aerosol particles is a sink for sulphur in the MBL (Sievering et al. 1992). More in general, heterogeneous reactions on aerosol surfaces has been invoked for their role in atmospheric chemistry (Dentener and Crutzen, 1993), but their effect on aerosol physical characteristics has not been studied.
Question 6. What are the rates and efficiencies of the processes involving MBL clouds that change the number size distribution and size-dependent chemical composition of a continentally derived aerosol as it advects over the North Atlantic? There are a variety of processes induced by clouds that can effect both the mass and number concentration of the aerosol. Major attention must still be given to the in-cloud oxidation of SO 2, which is the major pathway for producing sulfate mass in the global atmosphere (Hegg, 1985, Langner and Rodhe, 1991, Benkovitz et al., 1994). There are large uncertainties on the oxidation efficiencies under various conditions, which may account for a factor 1.5 of uncertainty on the estimates of the global direct forcing of sulphate aerosols (Kiehl and Rodhe, 1995). The cloud processing of gasses other than SO 2 is largely unknown. The role of drizzle and coagulation in fixing the levels of CCN at either a low or a high number concentration, depending on the formation rate of CCN (read: degree of pollution) has been discussed by Baker and Charslon, (1990), and Ackerman et al. (1994). These studies show that a source rate of a few times 10 -3 cm-3s -1 would be sufficient to sustain CCN number concentrations over the oceans.
Aspects of tropospheric aerosols: Questions and ACE-2
753
Such source rates can well be achieved by considering entrainment of particles from the free troposphere (Raes and Van Dingenen, 1996). Coalescence of water droplets, scavenging of interstitial aerosol by cloud droplets and precipitation play a major role in transforming the continental aerosol advecting within the MBL into a marine aerosol. Their combined rate of transformation is poorly quantified, but is needed to assess the geographical extent of the radiative effect of continental aerosols over the ocean. Falkowski et al. (1992) have argued that over the North Atlantic, the effect of anthropogenic pollution on cloud albedo would only be large close to the sources, and that further down-wind, the albedo would be governed by the natural, DMS derived aerosol. This may be in contrast with the measurements of Van Dingenen et al. (1995) who measured a correlation between the number of CCN and black carbon aerosols throughout the North Atlantic. Recent measurements have shown that clouds may also be a source of particles, through homogeneous nucleation that is promoted by high relative humidity in the vicinity of clouds (Hegg, et al., 1990, Perry and Hobbs, 1995, Wiedensohler, 1996). It is not known whether these particles eventually play a role in the budget of radiatively active particles or CCN. Progress has also been made in calculating the activating properties of internally mixed aerosols, such as sulfates with organic compounds (Shulman et al., 1996).
Question 7. How much do aerosols formed or transported in the free troposphere contribute to the direct radiative forcing over the North Atlantic, and how do they affect the properties of aerosols and clouds in the MBL? Long range transport in the stratified layers of the free troposphere has been documented for pollution outbreaks from Europe and North America (Parrish et al., 1994), for desert dust from Africa (Prospero and Carlson, 1972), and for biomass burning aerosols from Africa (Andreae et al., 1994). Sattelite pictures from S AGE-II document how during spring-summer the upper troposphere of the NHemishere becomes filled with aerosols, probably due to convective transfer of anthropogenic aerosols (Kent et a1.,1995). Obviously, these free tropospheric aerosol layers or plumes must be considered when assessing the direct effect of aerosols on radiative forcing. Aerosols in these plumes might also have an effect on MBL clouds through entrainment of the sub-micron aerosol or sedimentation of the larger particles. Free tropospheric particles that enter the MBL can take part in cloud droplet activation, depending on their size and chemical characteristics (Hudson, 1993). They also provide additional aerosol surface area on which H2SO4, MSA and low vapour pressure organic vapours, produced within the MBL, might condense rather than nucleate and form new particles. As such, entrainment of FT aerosols could determine the aerosol number concentration in the MBL, as well as the number of cloud droplets in MBL clouds (Raes, 1995). It has also been proposed that desert dust plumes might have an effect on the oxidizing capacity of the atmospere (Dentener et al., 1996). There exists experimental and theoretical evidence that in remote areas the background free tropospheric aerosol is sustained by in situ nucleation of H2SO4-H20 particles in the upper troposphere (Clarke, 1993, Raes et al, 1993). No measurements exist of nucleation events in the perturbed free troposphere over the North Atlantic. Measurements during ASTEX suggest that new particle formation by nucleation is quenched in FT plumes of continental origin, and that the aerosol number concentration is determined by refractory aerosols that initially advect from the continents (Clarke et al., 1995). The role of subsidence in the subtropics and its role as a mechanism for injecting aerosols in the surface layers must further be elucidated (Covert et al., 1996).
754
Raes
EXTRAPOLATION AND ASSESSMENT Question 8. How can locally derived aerosol characteristics and process information be extrapolated to regional and global scales? Increasing the accuracy of the present estimates of aerosol radiative forcing will eventually rely on the capabilities to use detailed aerosol characteristics and process information in global circulation models. Aerosol and cloud fields are inhomogeneous in time and space, and it is still a scientific problem how to extrapolate information obtained during an intensive campaign to the seasonal and regional scales which are of climatological relevance. Ideally, one needs a hierarchy of observations and models, from the local to the global scale, where the link between information at various levels of resolution is understood. In practice, this means (1) putting information obtained during an intensive campaign in the context of geographically widespread and long-term monitoring activities, (2) refming procedures that link in-situ observations with those obtained from satellites and (3) developing of model parametrizations. While satellites offer observations on a planetary scale, they have a reduced capability for chemical analysis of the aerosol particles and for coupling the geographical distribution of the aerosol to spatially and temporally variable aerosol precursors and source processes. Modelling of the aerosol system offers the only possibility for coupling locally measured properties to global forcing and effects. In particular, a close collaboration between the modelling and measuring community should assure that experimental results obtained during ACE-2 are processed by a hierarchy of models, and make an extrapolation of the results possible. This process must eventually lead to an assessment of radiative forcing by anthropogenic aerosols over the North Atlantic. CONCLUSION The ACE-2 steering committee has developed a strategy for tackling the questions discussed above. This strategy is based on closure experiments, experiments in a Lagrangian framework (Lenschow et al., 1981, Choularton et al., 1992), and a complementary deployment of measurements from land sites, ship, airplanes and satellites. The planning and development of the operations can be followed on the ACE-2 WWW home page: http://rea.ei.jrc.it/Nvandinge/ace2/ace2main.html. ACKNOWLEDGEMENTS With thanks to the ACE-2 Scientific Steering Committee and all ACE-2 investigators for stimulating discussions. At the moment of writing, ACE-2 has attracted funding from the European Commission's Environment & Climate Programme, UK Natural Environment Research Council, and the US National Science Foundation. REFERENCES Ackerman A.S., O.B. Toon, and P.V. Hobbs, Reassessing the dependence of cloud condensation nucleus concentration on formation rate, Nature, 367,445-447, 1994 Albrecht B.A., Aerosols, cloud microphysics, and fractional cloudiness, Science, 245, 1227-1230, 1989 Andreae M.O., B.E. Anderson, D.R. Blake, J.D. Bradshaw, J.E. Collins, G.L. Gregory, G.W. Sachse and M.C. Shipham, Influence of plumes from biomass burning on atmospheric chemistry over the equatorial and tropical South Atlantic during CITE 3, J. Geophys. Res., 99, 1279312808, 1994 Ayers G.P. and I.E. Galbally, A preliminary estimate of boundary layer-free troposphere entrainment velocity, Baseline 92, 10-15,1995 Baker M.B., and R.J. Charlson, Bistability of CCN concentrations and thermodynamics in the cloudtopped boundary layer, Nature, 345, 142-145, 1990
Aspects of tropospheric aerosols: Questions and ACE-2
755
Benkovitz C.M., C.M. Berkovitz, R.C. Easter, S. Nemesure, Richard Wagener, and S.E. Schwartz, Sulphate over the North Atlantic and adjacent continental regions: evaluation for october and november 1986 using a three-dimensional model driven observation-derived meteorology, J. Geophys. Res., 99, 20725-20756, 1994 Boucher O., and U. Lohman, The sulfate-CCN-cloud albedo effect: a sensitivity study with two general circulation models, Tellus, 47B, 281-300, 1995 Bretherton C.S., Austin P., and S.T. Siems., Cloudiness and Marine Boundary Layer Dynamics in the ASTEX Lagrangian Experiments. Part II: Cloudiness, Drizzle, Surface Fluxes and Entrainment, J. Atmos. Sci., 52, 2724-2735, 1995 Cess .R.D., M.H. Zhang, P. Minnis, L. Corsetti, E.G. Dutton, B.W. Forgan, D.P. Garber, W.L. Gates, J.J. Hack, E.F. Harrison, X.Jing, J.T. Kiehl, C.N. Long, J.-J. Morcrette, G.L. Potter, V. Ramanathan, B. Subasilar, C. H. Whilock, D.F. Young, Y. Zhou, Absorption of solar radiation by clouds: observations versus models, Science, 267,496- 499, 1995. Charlson R.J., J. Langner, H. Rodhe, C.B. Leovy, and S.G. Warren, Perturbation of the nothem hemisphere radiative balance by backscattering from anthropogenic sulfate aerosols Tellus, 43AB, 152-163, 1991 Charlson, R.J., J.E. Lovelock, M.O. Andreae and S,G. Warren, Oceanic phytoplankton, atmospheric sulfur, cloud albedo and climate, Nature, 326, 655-661 1987 Choularton T.W., R.M. Downer, B.J. Tyler, M.J. Gay, S.A. Penkett, B. Bandy, G.J., Dollard, T.J. Davies and B.M.R. Jones, A field study of the generation of nitrate in a hill cap cloud, Env. Pollution, 75, 69-73, 1992 Chylek P., G. Videen, D. Ngo, R.G. Pinnick, and J.D. Klett, Effect of black carbon on the optical properties and climate forcing of sulfate aerosols, J. Geophys. Res., 100, 16325-16332, 1995 Clarke A.D., Atmospheric nuclei in the Pacific midtroposphere: their nature, concentration, and evolution, J. Geophys. Res., 20633-20647, 1993 Clarke A.D., J.N. Porter, F.P.J. Valero, and P. Pilewskie, Vertical profiles, aerosol microphysics and optical closure during ASTEX: measured and modeled column optical properties, J. Geophys. Res. 101, 4443-4454,1996 Clarke A.D., T. Uehara, and J.N. Porter, Lagrangian evolution of an aerosol column during the Atlantic Stratocumulus Transition Experiment, J. Geophys. Res., 101,4351-4362, 1996 Coffman D.J., and D. Hegg, A preliminary study of the effect of ammonia on paticle nucleation in the marine boundary layer, J. Geophys. Res., 100, 7147-7160, 1995 Cooke W. and J. Wilson, A global black carbon aerosol model, J. Geophys. Res., (in print), 1996 Covert D.S., V.N. Kapustin, P.K. Quinn, and T.S. Bates, New particle formation in the marine boundary layer, J. Geophys. Res., 92, 20581-20589, 1992 Covert D.S., V.N. Kapustin, T.S. Bates, adn P.K. Quinn, Physical properties of marine boundary layer aerosol particles of the mid-Pacific in relation to sources and meteorological transport, J. Geophys. Res., 101, 6919-6930, 1996 Dentener F.J. and P.J. Crutzen, Reaction of N205 on tropospheric aerosols: impact on the global distributions of NOx, 03 and OH, J. Geophys. Res., 98, 7149-7163, 1993 Dentener F.J., G.R. Carmichael, and Y. Zhang, The role of mineral aerosol as a reactive surface in the global troposphere, J. Geophys. Res., (submitted), 1996 Falkowski P.G., Y. Kim, Z. Kolber, C. Wilson, C. Wirick, and R. Cess, Natural versus anthropogenic factors affecting low-level cloud albedo over the North Atlantic, Science, 256, 1311-1313, 1992 Haywood J.M. and K.P. Shine, The effect of anthropogenic sulfate and soot aerosol on the clear sky planetary radiation budget, Geophys. Res. Lett., 22, 603-606, 1995 Hegg D.A., The importance of liquid-phase oxidation od SO2 in the troposphere, J. Geophys. Res., 90, 3773-3779, 1985
756
Raes
Hegg, D.A., L.F. Radke and P.V. Hobbs, Particle production associated with marine clouds, J. Geophys. Res., 95, 13917-13925, 1990 Heintzenberg J., Light scattering parameters of internal and external mixtures of soot and nonabsorbing material in atmospheric aerosols, Conference on Carbonaceous Particles in the Atmosphere, Berkeley, CA, 278-281, 1978 Heintzenberg J., Fine particles in the global troposphere: a review, Tellus, 41 B, 149-160, 1989 Heintzenberg J., and D.S. Covert, On the distribution of physical and chemical particle properties in the atmospheric aerosol, J. Atmos. Chem., 10, 383-397, 1990 Hoppel, W.A., J.W. Fitzgerald, G.M. Frick and R.E. Larson, Aerosol size distributions and optical properties found in the marine boundary layer over the Atlantic Ocean, J. Geophys. Res, 95, 3659-3686, 1990 Husar R., and L.L. Stowe, Tropospheric aerosols over the oceans, http://capita. Wustl.edu/CAPITA/CapitaReports/TropoAerosol/trop0.html Hudson J.G., Cloud condensation nuclei near marine cumulus, J. Geophys. Res., 98, 2693-2702, 1993 Jones A., D.L. Roberts, and A. Slingo, A climate model study of indirect radiative forcing by anthropogenic sulphate aerosols Nature, 370, 450-453, 1994 Kerminen V.-M., A.S. Wexler and S. Potukuchi, Growth of freshly nucleated particles in the troposphere: roles of NH3, H2SO4, HNO3 and HC1, J. Geophys. Res. (submitted), 1996 Kent G.S, P.-H. Wang, M.P. McCormick and K.M. Skeens, Multilayer Stratospheric Aerosol and Gas Experiment II: measurements of upper tropospheric aerosol characteristics, J. Geophys. Res., 100, 13875-13899, 1995 Kiehl J.T., and H. Rodhe, Modelling geographical and seasonal forcing due to aerosols, in '"Aerosol forcong of climate" (Eds. R. Charlson and J. Heintzenberg), John Wiley, 281-296, 1995 Kiehl J.T., and B.P. Briegleb, The relative roles of sulfate aerosols and greenhouse gases in climate forcing, Science, 260, 311-314, 1993 Langner, J., and H. Rodhe, A global three-dimensional model of the tropospheric sulfur cycle, J. Atmos. Chem., 13, 225-263, 1991 Lenschow, D. H., R. Pearson, Jr., and B. B. Stankov, Estimating the ozone budget in the boundary layer by use of aircraft measurements of ozone eddy flux and mean concentration, J. Geophys. Res., 86, 7291-7297, 1981 Li X., H. Maring, D. Savoie, J. Prospero, Dominance of mineral dust in aerosol light-scattering in the North Atlantic trade winds, Nature, 380, 416-419, 1996 Martin G.M., D.W. Johnson and A. Spice, The measurement and parametrization of effective radius of droplets in warm stratocumulus clouds. J. Atmos. Sci., 51 13 1823-1842, 1994 McGovern F.M., S.G. Jennings, T.C. O'Connor and P.G. Simmonds, Aerosol and trace gas measurements during the Mace Head Experiment, Atmos. Env. (in print), 1996 O'Dowd, C.D. and M.H. Smith, Physicochemical properties of aerosols over the Northeast Atlantic: Evidence for wind-speed-related submicron sea-salt aerosol production, J. Geophys. Res.,98, 1137-1149, 1993 Pandis S.N., L.M. Russell, and J.H. Seinfeld, The relationship between DMS flux and CCN concentration in remote marine regions, J. Geophys. Res., 99, 16945, 1994 Parrish D.D., J.S. Holloway, M. Trainer, P.C. Murphy, G.L. Forbes, and F.C. Fehsenfeld, Export of North American ozone pollution, Science (in press), 1994 Perry K.D., and P.V. Hobbs, Further evidence for particle nucleation in clean air adjacent to marine cumulus clouds, J. Geophys. Res., 99, 22803-22818, 1994 Pham M., J.-F. Muller, G.P. Brasseur, C. Granier, and G. Megie, A three-dimensional study of the tropospheric sulfur cycle, J. Geophys. Res., 100, 26061-26092, 1995 Pilinis C., Pandis S., and J.H. Seinfeld, Sensitivity of climate forcing by atmospheric aerosols to aerosol size av,d composition, J. Geophys. Res., 100, 18739-18754, 1995
Aspects of tropospheric aerosols: Questions and ACE-2
757
Pincus R. and M.B. Baker, Effects of precipitation on the albedo susceptibility of clouds in the marine boundary layer, Nature, 372, 250-252, 1994 Prosper o J.M. and T.N. Carlson, Vertical distribution of Saharan dust over the western equatorial North Atlantic Ocean, J. Geophys. Res, 77, 5255-5265, 1972 Prospero J.M., R. Schmitt, E. Cuevas, D.L. Savoie, W.C. Graustein, K.K. Turekian, A. Volz-Thomas, A. Diaz, S.J. Oltmans, H.Levy II, Temporal variability of summer-time ozone and aerosols in the free troposphere over the eastern North Atlantic, Geophys. Res. Lett., 22, 2925-2928, 1995 Putaud J.P., R. Van Dingenen, F. Raes, F. McGovern, M. Mangoni, J. Prospero, and H. Maring, Results of the pre-ACE2 campaigns in the sub-tropical North Atlanitic, this issue, 1996. Quinn P., S.F. Marshall, T.S. Bates, D.S. Covert, and V.N. Kapustin, Comparison of measured and calculated aerosol properties relevant to the direct radiative forcing of tropospheric sulphate aerosol on climate, J. Geophys. Res.,(submitted) 1994 Raes, F., R. Van Dingenen, J. Wilson, and A. Saltelli, Cloud condensation nuclei from dimethyl sulphide in the natural marine boundary layer: remote vs. in-situ production, in 'DMS: Ocean, Atmosphere and Climate (G. Restelli and G. Angeletti, eds), Kluwer Academic Publisher, Dordrecht, 311-322, 1993 Raes F., J. Wilson, and R. Van Dingenen, Aerosl dynamics and its implication for the global aerosol climatology, in "Aerosol Forcing of Climate", (Eds. R.J. Charson and J. Heintzenberg) , John Wiley & Sons, 153 - 170 1995 Raes, F., Entrainment of free tropospheric aerosols as a regulating mechanism for cloud condensation nuclei in the remote marine boundary layer, J. Geophys. Res. ,100, 2893-2903, 1995 Raes, F. and R. Van Dingenen, Comment on "The relationship between DMS flux and CCN concentrations in remote marine regions" by S.N. Pandis, L.M. Russell, and J.H. Seinfeld, J. Geophys. Res., 100, 14355-14356, 1995 Raes, F.R. and R. Van Dingenen, Simultaneous measurements of aerosols in the free troposphere and marine boundary layer: evidence for a dominant free tropospheric source of marine boundary ayer CN and CCN, Geophys. Res. Lett. (submitted), 1996 Russell P.B., S.A. Kinne, and R.W. Bergstrom, Aerosol climate effects: local radiative forcing and column closure experiments, (in preparation). Schult I. and P. Koepke, Aerosol forcing based on Global Aerosol Data Set, 4th International Aerosol Conference, August 29 - September 2, Los Angeles, 1994. Saxena P., L.M. Hildemann, P.H. McMurry, and J. Seinfeld, Organics alter hygroscopic behaviour of atmospheric particles, J. Geophys. Res., 100, 18755-18770, 1995 Shulman M.L., M.J. Jacobson, R.J. Charlson, R.E. Synovec, and T.E. Young, Dissolution behaviour and surface tension effects of organic compounds in nucleating cloud droplets, Geophys. Res. Lett, 23,277-280, 1996 Sievering H., J. Boatman, E. Gorman, Y. Kim, L. Anderson, G. Ennis, M. Luria, and S. Pandis, Removal of sulfur from the marine boundary layer by ozone oxidation in sea-salt aerosols, Nature, 360, 571-573, 1992 Strom J, K. Okada, and J. Heintzenberg On the state of mixing of particles due to Brownian coagulation, J. Aerosol Sci., 23,467-480 1992 Svenningsson I.B., H.-C. Hansson, A. Wiedensohler, J.A. Ogren, K.J. Noone, and A. Hallberg, Hygroscopic growth of aerosol in the Po Valley, Tellus, 44B, 556-569, 1992 Taylor K.E., and J. Penner, Response of the climate system to atmospheric aerosols and greenhouse gases, Nature, 369, 734-737, 1994 Tegen I., and I. Fung, Modeling of mineral dust in the atmosphere: Sources, transport, and optical thickness, J. Geophys. R., 99, 22897-22914. Tegen I., A.A. Lacis, and I. Fung, The influence on climate forcing of mineral aerosols from disturbed soils, Nature, 380, 419-422, 1996
758
Raes
Twomey, S.A., M. Piepgrass, and T.L. Wolfe, An assessment of the impact of pollution on global cloud albedo, Tellus, 36B, 356-366, 1984 Van Dingenen, R., J. Hjorth and, F. Raes, Evidence for anthropogenic impact on number concentration and sulphate content of cloud-processed aerosol particles over the North-Atlantic, J. Geophys. Res., 100, 21057-21067, 1995 Van Dingenen R., M. Mangoni, J.-P. Putaud, and U. Watjen, Ultrafine number size distribution measurements and chemical characterisation of the aerosol over the Atlantic Ocean between 50 degN and 50 degS., this issue. Wiedensohler A., H.-C. Hansson, D. Orsini, M. Wendish, F. Wagner, K. Bower, T.W. Choularton, M. Wells, M. Parkin, K. Acker, W. Wieprecht, M. Facchini, J.A. Lind, S. Fuzzi, B.G. Arends, M. Kulmala, Night-time formation and occurrence of new particles associated with orographic clouds (in preparation) Wiscombe W.J., An absorbing mistery, Nature, 376, 466-467, 1995
THE CONTRIBUTION OF CARBONACEOUS
AEROSOLS
TO
CLIMATE
CHANGE
Joyce E. Penner 1,2, Catherine C. Chuang 1, and Catherine Liousse 3 IAtmospheric Science Division, Lawrence Livermore National Laboratory, Livermore, CA 2Department of Atmospheric, Oceanic, and Space Sciences, University of Michigan, Ann Arbor, MI 3Centre des Faibles Radioactivit6s, CNRS-CEA, ave de la terrasse, 91198 Gif sur Yvette, France Abstract - The contribution of aerosols to climate change results from two effects: clearsky and cloudy-sky forcing. The clear-sky climate forcing by carbonaceous aerosols from biomass burning and fossil fuel burning depends on the relative contribution of scattering and absorption by the aerosols which in turn depends on the fraction of aerosol mass associated with black carbon and its size distribution. In this paper, we review estimates for the emission of carbonaceous aerosols, placing these estimates in the context of estimates for the emissions of anthropogenic and natural sulfate aerosols and natural sources of organic particulate matter. The cloudy-sky forcing from carbonaceous aerosols is difficult to estimate because, among other factors, it depends on the amount of absorption by the aerosols in the cloud. It is also highly sensitive to the assumed pre-existing, natural aerosol abundance. An upper limit for the cloudy-sky forcing by carbonaceous aerosols is -4.4 Wm -2, but may range as low as -2.4 Wm -2, depending on background aerosol concentrations. These estimates do not yet account for absorption of radiation by black carbon associated with cloud or the presence of pre-existing dust particles. Keywords - Radiative forcing; Aerosols; Cloud droplet nucleation; Climate change INTRODUCTION Atmospheric aerosols along with greenhouse gases and clouds play important roles in mediating the radiation balance of the Earth-atmosphere system. In recent assessments of climate forcing, the Intergovernmental Panel on Climate Change (IPCC) lists aerosols as one of the most important anthropogenic radiative agents that tend to decrease temperature (IPCC, 1994). However, the magnitude of aerosol effects on climate and, in particular, the magnitude of the indirect effect of aerosols on clouds is still very uncertain (IPCC, 1994; Penner et al., 1994a). While a variety of aerosol types exist in the atmosphere (e.g., water-soluble inorganic species, carbonaceous aerosols (i.e. organic matter and black carbon aerosols), mineral dust, and sea salt), volatile sulfur compounds are particularly important aerosol precursors. This is because a large part of the aerosol mass in sub-micron size particles is sulfate. Sulfate aerosol is formed through chemical reactions, either gas phase photochemical reactions of the emitted sulfur compounds or aqueous processes involving in-cloud oxidation of SO2 followed by drop evaporation (Hoppel et al., 1990; Lelieveld and Heintzenberg, 1992). The second most important aerosol component in the sub-micron size range is organic carbon. Our studies of the effect of aerosols on climate and climate forcing have initially concentrated on characterizing the effects of anthropogenic sulfate aerosols (Penner et al., 1994b; Taylor and Penner, 1994; Chuang and Penner, 1995; Chuang et al., 1995). Here we review these estimates of forcing and extend them to include an upper limit for the cloudy-sky forcing by carbonaceous aerosols. In order to provide a global understanding of the effects of aerosols on clouds, one must first understand the global concentrations of the different aerosol components or types. Our past work has been aimed at developing an understanding of global and regional aerosol abundances, and developing a parameterization of cloud response to aerosol abundance (Chuang and Penner, 1995; Ghan et al., 1993) in order to evaluate the importance of aerosol/cloud interactions to climate forcing. The effect of aerosol abundance on cloud life cycle may also be important but is not treated here. In order to understand whether the aerosol particles act as a CCN, one needs to know the composition of hygroscopic material in the aerosol (e.g., sulfate, nitrates, ammonium) (Pruppacher and Klett, 1978). This understanding requires a quantitative understanding, on a global basis, of the aerosol sources, transformation and
759
760
Penner et al.
removal processes. Determining the aerosol sources requires an understanding of the cycles and budgets for the trace species which comprise the aerosol. Carbonaceous aerosols derive from both anthropogenic and natural sources. They are composed of two components, called black carbon and organic carbon. Organic carbon, like sulfate aerosol is mainly scattering. Black carbon may be distinguished by its resistence to chemical and thermal attack and by its ability to strongly absorb solar radiation; specific absorption coefficients are estimated in the range of 3 to 20 m2g (Liousse et al., 1993). This ability lowers the single scattering albedo of aerosols thereby -reducing the amount of solar radiation reflected by the aerosols (Chylek and Coakley, 1974). It is important to quantify the amount of black carbon (as well as organic carbon and sulfate aerosol) in order to quantify climate forcing by anthropogenic aerosols. The presence of black carbon in cloud may also reduce cloud albedo. Here we ignore this effect and concentrate on obtaining an upper-limit estimate of the effect of carbonaceous aerosols on clouds and on showing the sensitivity of this forcing to preexisting aerosol number concentrations. SOURCES OF CARBONACEOUS AEROSOLS The principal source of black carbon is combustion. Except for natural fires, whose sources are small on a global basis, most black carbon derives from either biomass burning or fossil fuel combustion. The principal types of biomass burning include (1) savanna fires to clear and renew land, (2) forest fires for clearing purposes, (3) burning of agricultural waste to clear land, and (4) the burning of wood to produce charcoal and (5) the burning of wood, agricultural wastes, charcoal, and dung for domestic fuel. Each of these processes produce organic carbon as well, although the ratio of BC to OC in emissions varies depending on the type of fuel and the manner of burning. For example, savanna fires typically have a larger ratio of BC/OC than forest fires. This is because savanna fires typically burn in a flaming mode which enhances emissions of BC while forest fires are a combination of flaming and smoldering. The principal sources of black carbon from fossil fuel emissions derive from diesel fuel use and coal combustion (Penner et al., 1993; Cooke and Wilson, 1995). On a global basis these emissions have been estimated from fuel use statistics as 6.6 Tg/yr by Penner et al. (1993) and as 8.0 Tg/yr by Cooke and Wilson (1995). In the following estimates of forcing, the inventory of Penner et al. (1993) has been used. Emissions of organic carbon from fossil fuels and other activities within an urban complex are not well known, but, typically, the concentration of OC is larger than that of BC in urban areas (Bremond et al., 1989; Rogge et al., 1993). The source of this OC is not necessarily associated with the main sources of BC in any given urban area (Gray et al., 1984). In the absence of good emissions inventories for OC from these sources, in the following we used an estimate for the ratio of OC to BC emissions from measured ratios of OC to BC in urban regions (which range up to 3.5 gC/gC) to define the range of possible emissions of OC from fossil fuel burning and urban activities. The resulting total emissions must also account for the particular molecular form of the organic compounds measured as OC. Assuming a conversion factor of 1.3 gives an estimated upper limit for these emissions of approximately 30 Tg/yr. We have distributed these emissions proportional to our emissions of BC from fossil fuels, but we stress here the uncertainty in this distribution as well as its magnitude. It is important to obtain a better understanding of these emissions and their distribution as a function of space and time in order to improve estimates of forcing by carbonaceous aerosols. In order to estimate forcing by carbonaceous aerosols, we developed a detailed inventory of anthropogenic emissions from biomass burning. For this inventory we estimated the fraction of BC and OC emitted from each source type using measured values of emission factors from the literature. For this inventory, forest and savanna burning were taken from the work of Hao et al. (1990), but updated to account for new information on the frequency of burning and the carbon stocks available to burn in grasslands (Menaut et al., 1991). Agricultural burning was estimated from country by country totals for the production of rice, wheat, barley, oat, rye and course grains, corn, and sugar cane as derived from FAO (1991). The total burned took into account the fraction used for fuel (-60% in developing countries; Middleton and Darley, 1973; Mahtab and Islam, 1984; Barnard and Kristofersen, 1985; Crutzen and Andreae, 1990) and the fraction burned in the field. For the latter calculation we estimated the total amount of agricultural waste associated with the amount of grain produced from the harvest index of each crop. The fraction of this waste submitted to fires and the combustion efficiency (or fraction of waste submitted to fires that actually burns) were estimated separately. Table 1 gives a summary of the factors which contribute to our calculations and emissions.
The contribution of carbonaceous aerosols to climate change
761
Table 1. Biomass fires : Coefficients used to obtain amount of burned biomass from crop and fuels. By Product I / By Product Combustion Emission Emission Main Product Exposed to Efficiency (%) Factor Factor Fires (%) TP (~/k~ fuel) BC (~/k~ fuel) Savanna 8.1 0.81 Forest 18 1.53 Rice Developed Areas 1.2 10 85 2.5 .6 Developing Areas 1.2 50 85 5.32 .8642 Wheat, Barley, Rye and others Developed Areas 1.3 5 70 7.0 .91 Developing Areas 1.3 30 70 12.02 1.22 Corn Developed Areas 1.0 5 35 5.0 .7 Developing Areas 1.85 30 35 12.02 .962
Sugar Cane Developed Areas Developing Areas Wood Developed Areas Developing Areas Burned for fuel Burned for charcoal making Charcoal Developed Areas Developing Areas Dung
0.5 0.5
10 70
35 35
5.0 7.02
.775 .772
80
90
11
1.32
50
90
11
1.2
50
20
17
0.85
100 100 18.5
75 75 50
10 10 20
1.5 1.5 1.0
l In the case of rice, wheat, rye, barley and corn, the main product is the grain and the by product (or waste), the straw. Sugar is considered to be the main product of the sugar cane crop. 2These emission factors are higher because 60% of the burning is assumed to take place as domestic burning rather than in the field. There is consequently a higher fraction of smoldering with higher total particle emission factors but a lower fraction of BC/TP emissions. In addition to forest, savanna, and agricultural burning (in both field and as domestic fuel), charcoal, wood, and dung are burned as fuels. The total amounts burned from these sources and the total emissions are shown in Table 2. There is a crucial lack of data concerning the proper parameters to use in these calculations. For example, the emission factors for charcoal making and burning of wood directly differ greatly. In our inventory, 50% of the fuel wood was estimated to be used for charcoal making in the developing countries and 50% for direct fuel use. This ratio could be improved based on recent evaluations (Brocard et al., 1995). To check the magnitude and distribution of the emissions described above, requires that we represent these aerosols in a three-dimensional model of transport, transformation and removal and use the predictions of the model in comparison to observations to quantify possible errors in both emissions and model representation. Black carbon concentrations are almost exclusively from combustion and of anthropogenic origin, while measured abundances of OC at any given site also result from natural sources. Therefore, a proper test of the model and the emissions of OC, requires that both sources be represented. Natural emissions of OC derive from the gas to particle conversion of natural organic vapors emitted by vegetation, direct particle emissions from vegetation and other debri of biological origin. Vegetation emits isoprene, terpenes, and a variety of oxygenated species. All of these compounds
762
Penner et al.
Table 2. Biomass Burning Sources" Total fuel burned in each category (savanna, forest, agricultural and domestic fires). BC Emissions Total Biomass TP Emissions (Y~/yr) Burned (Y~dm/yr) (Y~/~'r) 5.64 5374.5 59.9 Total Savanna 2682 21.7 2.2 Forest 1259 22.7 1.9 4.45 564 0.53 Agricultural 192 2.06 0.22 Fires 0.46 41.1 0.04 Wheat and other grains 1.15 0.19 218.5 Corn 112.4 Rice 0.78 0.09 Sugar Cane 11.07 Domestic 869.5 1.01 Fires 752.4 8.88 0.88 0.16 0.02 Wood and Bagasse 16 Charcoal 101.1 2.02 0.10 Dung undergo photochemical degradation in the atmosphere, but terpenes are thought to have the largest particle yields. The fraction of organic matter produced from a given amount of terpene emissions varies over a considerable range, but a gas to particle conversion factor of 5% represents a reasonable average (Pandis, et al., 1991; Hatakeyama et al., 1991; Zhang et al., 1992). This factor, together with the terpene emission inventory of Guenther et al. (1995) gives rise to a total natural organic matter production of 7.8 Tg/yr. Because the time scale for oxidation of the terpenes is short (of the order of minutes), we assume that this source is directly injected into the model as aerosol. The latter two natural sources, those from direct injection of plant materials and biological debri, are particularly difficult to estimate. However, most of the mass of these emissions are in the coarse particle mode, above 1 lam diameter. Hence, their contribution to measured concentrations < 2.5 lam should not be large. In the following simulations, these emissions have been neglected. COMPARISON TO DATA There are very few data which can be used to verify the model, and much of it consists of short duration measurements which are therefore insufficient for defining a climatology of carbonaceous aerosols (especially a climatology of organic carbon aerosols). Liousse, et al. (1995) have carried out a thorough comparison of the model with those measurements which are available, concluding that the model provides a reasonable description of both OC and BC concentrations. Figure lab shows a scatter plot of observed and model predicted concentrations of BC and OC. The figure demonstrates that the emissions and model are not biased either high or low, although, if we look in more detail, particularly in the United States, for example, concentrations of both BC and OC appear to be low (Liousse, et al., 1995). This is most likely the result of the emission inventory being too low in the United States rather than due to weaknesses in the model. One weakness of the model is apparent when comparing modeled and predicted concentrations at polar locales. The predicted concentrations at both the North and South Poles appear to be too low. Low concentrations at remote locales may result from either (1) low emissions inventory, (2) a scavenging parameterization which removes too much material, or (3) a poor representation of atmospheric circulation and transport. The representation of scavenging may be evaluated by comparison of predicted and observed concentrations in precipitation (see Liousse et al., 1995) or by comparison of observations and model results at remote sites, where predicted concentrations are highly sensitive to the scavenging parameterization. Figure 2 compares the model and predicted concentrations of BC at Mauna Loa, a site mainly impacted by emissions in Asia. As seen there, the predicted concentrations are reproduced to within a factor of two. One complicating factor in comparing these data is the fact that BC as measured by an aethelometer may count some absorption by dust, attributing this to black carbon. Schnell et al. (1994) presented evidence that dust absorption could account for up to 20% of the BC measured at Mauna Loa during special Asian dust events (mean value of 10%). Our predicted concentrations are therefore within a factor of two of those measured which we deem adequate for this first estimate. This fact, together with
The contribution of carbonaceous aerosols to climate change
763
our reasonable reproduction of concentrations in rain (Liousse et al., 1995) gives us confidence in our scavenging scheme. South Polar concentrations are affected mainly by biomass burning in the Southern Hemisphere. The magnitude of these sources is best checked by comparison of the predicted concentrations with sites close to the burning. One such comparison, for Amsterdam Island, gave good agreement, thereby confirming that our sources from Africa are reasonable (Liousse et al., 1995). The seasonal cycle of carbon monoxide is also mainly dependent on the sources from biomass burning in the Southern Hemisphere and is a gas whose sink does not depend on the scavenging parameterization in the model. A comparison of this model's predicted CO fields (see Atherton et al., 1995) with data from Cape Grim (not shown) gives us further confidence that the specified emissions from biomass burning are reasonable. Hence, we conclude that the low concentrations predicted at the South Pole are mainly due to a poor representation of the circulation in this region. A similar weakness may affect the model's ability to represent concentrations in North Polar regions. This weakness should not greatly affect the predicted forcing by carbonaceous aerosols, because such a small region is affected. However, we hope to correct this in the future.
(a)
300
(b)
2000 "
2OO
.fi v
n~
100'
0
0
g o
/l~. ~ ~r
)0
go o oo
9
~ 9
100 2~)0 Observed (BC) in ng/m 3
0
300
oo
~
o
o
o o
0
10~30 Observed (OC) in ng/m 3
Figure 1 ab: Predicted and observed concentrations of black carbon (a) and organic matter (b).
-
30
&~'~~m20
o
!
.
.
- r
-
1991
Experimental data *........=__1992 ~ 1993 Predicted by model
t t ~ /~ / \ '~
.
.
.
In Fb Mr Ap My Je *Bodhaine et al., 1995
II
Ag Sp Ot
Nv l~
Figure 2: Predicted and observed concentrations of black carbon at Mauna Loa.
2000
764
Penner et al.
THE EFFECTS OF AEROSOLS ON CLOUDS Recently, Boucher and Rodhe (1994) and Jones et al. (1994) have each developed parameterizations relating cloud drop concentration to sulfate mass or aerosol number concentration, respectively, and used them to develop estimates of the indirect forcing by anthropogenic sulfate aerosols. These parameterizations made use of measured relationships in continental and maritime clouds. However, these relationships are inherently noisy, yielding more than a factor of 2 variation in cloud drop number concentration for a given aerosol number ( or for a given sulfate mass) concentration. They do not make use of information from the climate model regarding local updraft velocities, and they have had to make certain simplifying assumptions. In the study by Boucher and Rodhe (1994), simultaneous aerosol mass and CCN number concentration measurements were used to establish a hypothetical range of dependencies between the sulfate aerosol mass and cloud drop number concentration. However, their results of the indirect radiative forcing by sulfate aerosols are sensitive to these assumed relationships and the predicted forcing may be overestimated when comparing with observations of cloud drop radius and cloud albedo. In the study by Jones et al. (1994), aerosol number concentration was related to cloud drop number concentration based on measurements. But in converting the model-predicted sulfate mass concentrations to aerosol number concentrations for use in the parameterization, the aerosol composition and size distribution were prescribed and assumed to be universally applicable both in the case of the natural background aerosol (present prior to industrialization) and in the present-day perturbed case. In contrast to previous studies, our parameterization of the effects of aerosols on cloud droplet distributions uses a more mechanistic approach. The characteristics of the cloud drop size distribution near cloud base are initially determined by the size distribution and chemical characteristics of the aerosol particles that serve as CCN and by the local updraft velocity (Lee et al., 1980; Chuang et al., 1992). Once drop concentrations at cloud base are established, measurements have shown that these remain near constant with altitude throughout the main part of the clouds, at least in the case of stratiform and stratocumulus clouds (Nicholls, 1984; Bower et al., 1994; Mitchell, DRI, private communication). Thus, ideally, it should be possible to use these fundamental properties of aerosol size, chemical composition, and updraft at cloud base to predict the effects of anthropogenic aerosols on drop number concentrations in stratiform clouds in general circulation models. However, the large spatial and temporal variability in the concentration, chemical characteristics, and size distribution of aerosols have made it difficult to develop such a parameterization from data. In our parameterization, our focus has been to develop a means for relating the predicted anthropogenic sulfate mass to cloud drop number concentration over the range of expected conditions associated with continental and marine aerosols. We start with an assumed pre-existing particle size distribution and develop an approximation of the altered distribution after addition of anthropogenic sulfate. This treatment is necessary for sulfate aerosols because the time scale for production from gas phase SO2 is several days. Other aerosol types, such as the carbonaceous particles of concern here, are formed much more quickly from their gas phase precursors and hence may be assumed to be injected into a global-scale model in the aerosol form. In the calculation of sulfate aerosol, we assume that sulfate is formed mainly through the condensation of sulfuric acid vapor (H2SO4) on pre-existing particles and aqueous-phase oxidation of SO2 followed by evaporation of the drops. The nucleation of new particles is neglected. We thereby develop a conservative estimate of the possible change in CCN number concentration due to anthropogenic sulfate-containing aerosols and investigate its impact on cloud albedo and global average solar radiative forcing. As noted above, the assumed pre-existing particle size distribution derives from a variety of sources, including the gas-to-particle conversion of natural emissions of non-methane hydrocarbons (primarily terpenes), natural water-soluble inorganic species (sulfates, nitrates, ammonium), dust, and sea salt. The sources of these aerosols are regionally diverse. We are working towards the development of global scale distributions for each of these aerosol types. At the present time, we are able to represent the most important sub-micron components which comprise the main mass of sub-micron aerosol, namely organic, black carbon, and sulfate aerosols (Liousse et al., 1995; Penner et al., 1994b). However, the pre-existing aerosol should also include a representation of particles whose mass is mainly in the course mode as well. In Chuang et al. (1995) and here, we assume the sulfate-containing aerosol is an internal mixture, where a fraction of the aerosol size distribution is determined by condensation of sulfuric acid vapor on a prescribed pre-existing particle distribution and a fraction was determined by aqueous-phase oxidation of
The contribution of carbonaceous aerosols to climate change
765
802 followed by evaporation of the drops. Based on Langner and Rhode (1991), as well as our own model simulations (Penner et al., 1994c), we explored a range of conditions for the gas and aqueous production pathways. Thus, it was assumed that 15 to 35% of aerosol sulfate is distributed to all particles by condensation and the remaining 85 to 65% to particles which were larger than the minimum size of CCN by cloud processes. Here, we use the mid-range figure of 75% to calculated the cloud forcing by anthropogenic sulfate aerosol. This internal mixing approach does not change the total aerosol number, but the resulting sulfate-containing particles grow to larger sizes and thereby form more CCN. CLOUD DROP NUCLEATION An aerosol particle becomes activated as a CCN when the environmental supersaturation ratio becomes greater than its critical value. The resulting drop grows to a size much larger than the initial size of the particle due to condensation of water. The magnitude of the critical supersaturation for activation decreases with increasing particle size and mass fraction of soluble material in the aerosol particle; therefore, larger particles are activated earlier than smaller particles and activation of maritime aerosols whose composition contains a large fraction of soluble salts (Heintzenberg, 1989) is more favorable than continental particles for a given value of supersaturation. Here, we describe our parameterization to relate anthropogenic sulfate-containing aerosols to cloud drop nucleation. The detailed microphysical model described by Chuang et al. (1992) was used to compute the spectral evolution of interstitial aerosols and cloud drops. This model describes a Lagrangian air parcel which may also entrain environmental air. The model is initialized with the aerosol size distribution and chemical composition determined as noted above and computes the concentration of cloud droplets formed as the parcel passes through cloud base. In Chuang et al. (1995) and Chuang and Penner (1995), we assumed the parcel dynamics were adiabatic because we were interested only in the initial stages of cloud development. The vertical temperature and humidity profiles were those used by Lee et al. (1980). We examined the consequence of water vapor diffusion to aerosols which were an internal mixture of sulfate and other materials (organic matter, NO3-, NH4+,. etc.). Figures 3a and 3b present the predicted relationship between anthropogenic sulfate in particles and cloud drop number concentration as predicted by our microphysical model. In these figures, we varied the aerosol number concentration from 500 - 10000 cm -3 for the continental case (Fig. 3a) and from 50 500 cm -3 for the marine case (Fig. 3b) to cover the spatial and temporal variations expected for the concentrations of pre-existing aerosols. In addition, we also considered updraft velocities ranging from 10 cm s -1 to 200 cm s -1 to cover a wide range of updrafts expected in both stratus and stratocumulus clouds. Looking at any one curve in Fig. 3, one notes that under many circumstances the predicted cloud drop number concentration is larger on particle size distributions with a higher loading of anthropogenic sulfate even though, for this internally mixed case, the total number of aerosol particles is unchanged in these simulations. The concentration of cloud drops is, of course, also affected by both updraft velocity and total (in this case pre-existing) aerosol number. For the continental case with a moderate updraft velocity 50 cm s -1 and aerosol number 2000 cm -3, we estimate that the number of the condensationallyproduced cloud drops would increase from 350 to 560 cm -3 if a typical amount of anthropogenic sulfate o f - 2 lag m -3 in non-urban regions were added onto the pre-existing particles. For the marine case with the same updraft velocity but a much lower aerosol concentration of only 100 cm -3, the number of cloud drops nucleated would increase from 60 to 75 cm -3 for a total deposition of anthropogenic sulfate equal to 0.2 ~tg m -3. It is important to try to compare these results with available data. We have compared our model results with measurements of the relationship between drop number concentration and sulfate concentration in cloud water samples (Leaitch et al., 1992). Leaitch et al. report measurements conducted over central Ontario, Canada and over upper New York State. The observed ranges of cloud drop number concentrations are about 55 - 750 cm -3 for stratiform clouds and 160 - 950 cm -3 for cumuliform clouds (derived from Eqs. (1) and (2) of Leaitch et al., 1992) for cloudwater sulfate concentrations ranging from 0.05 to 10 gg m -3. For the same range of sulfate concentrations (presumably anthropogenic), our simulations show that the number concentrations of cloud drops which are nucleated on continental aerosol size distributions with total number concentrations 500 - 10000 cm -3 are about 50 - 910 cm -3 for updraft velocities 10 - 50 cm s -l. The predicted cloud drop numbers through nucleation on anthropogenic sulfate-containing aerosols seem to be in the right range.
Penner et al.
766 w = 10 c m / s 10"
.
.
w = 50
,
104
-
w = 200 cm/s
cm/s
.
. . . . . .
,
104
.
.
.
.
|
-
I
.
.
.
N = 500 cm-3 N = 2000 .....
N=
I0000
10~
10 3
10:
J
10~
10 ~ l.t
10 J 0
.
.
.
.
.
.
.
.
101
.
5
,
,
.
:
0
10
I
.
.
.
101
.
5
10
.
.
0
.
.
.
10
5
anthropogenic sulfate (gg m-3)
Figure 3a: Predicted cloud drop number nucleated versus anthropogenic sulfate in continental particles. Aerosol number concentrations are from 500 -10000 cm -3. w = 200 ends w = 50 cm/s w = 10 cngs 103 10 2 . . . . . . . . . 10 3 9
,
,
,
!
9
,
,
,
i
i
i
i
,
,
i
i
i
!
i
i
,
,
.
.
N = 50 cm-3 N=I00 N=500
......
l,t
10 ~
.
102
10 2 /
@
10 j
0
i
,
,
9
0
~
.
.
.
.
1.0
10 ~
0
,
9
,
,
,
0.5
.
.
.
.
.
.
1.0 1010
0.5
1.0
anthropogenic sulfate (~tg m-3)
Figure 3b. Same as Fig. 3a but for marine aerosols with number concentrations from 50 - 500 cm -3. PARAMETERIZATION AND FORCING IN A GLOBAL MODEL The effect of anthropogenic sulfate accumulation on pre-existing aerosols has a significant influence on cloud drop number concentration; therefore, it should be represented in climate models. However, it is not practical to apply a detailed microphysical model in a global climate model because of the large increase in computational time required for the detailed treatment of nucleation at each grid point. An alternative is to parameterize the cloud drop nucleation and then apply this parameterization in climate models. Here, we parameterize the nucleation of cloud drops onto anthropogenic sulfate-containing aerosols based on the results from our detailed microphysical model. The cloud drop number nucleated (in cm -3) is expressed in the form Ncl = w Na / (w + c Na) as suggested by Ghan et al. (1993), where Na (cm -3) is the background aerosol number concentration, and w is the updraft velocity in cm s-1. In Ghan et al., a look-up table was used for scaling the value of c such that the parameterized drop concentrations match the simulated values from our detailed microphysical model (Chuang et al., 1992; Edwards and Penner, 1988). In Chuang and Penner (1995) and Chuang et al. (1995), we derive the coefficient c directly from the results of our microphysical model.
The contribution of carbonaceous aerosols to climate change
767
Table 3. Global average annual mean change in cloud forcing (Wm-2).
Aerosol type Anthropogenic sulfate Anthropogenic carbonaceous aerosol Anthropogenic sulfate and anthropogenic carbonaceous aerosol
Forcing (Wm -2)
Forcing with prescribed preexisting marine source (Wm -2)
Anthropogenic OC and BC, Natural OC, BC, S
-0.87
-0.57
Natural OC, BC, S
-4.41
-2.43
Natural OC, BC, S
-5.29
Assumed preexisting aerosol
Results from this parameterization for sulfate cloud forcing using all carbonaceous aerosols as pre-existing aerosols are shown in Table 3. The total forcing noted there is --0.9 Wm -2- This table shows the predicted forcing from the model which assumes internal mixing and an assumed conversion fraction of 75% from aqueous reactions. We also show the predicted forcing for a calculation in which the preexisting aerosol is assumed to be only that from natural sulfur and natural organic matter aerosol. Here, the size distribution is that of the pre-existing aerosols from Chuang et al. (1995) and the anthropogenic carbonaceous aerosols have been added as an external mixture. The calculated forcing increases to 5.5Wm -2 with approximately-4.4 Wm -2 due to carbonaceous aerosols alone. Such forcing should be interpreted as a maximum since we have assumed that any effect of black carbon on cloud drop reflectivity is negligible. In addition, we have not yet accounted for the "pre-existing" aerosol component that arises from sea salt or dust. In order to explore the possible effects of the latter, we performed a sensitivity test with a prescribed background of aerosol over marine areas. In this test, a prescribed particle number concentration of 200 cm -3 was added to the lowest marine boundary layer, and exponentially decreased with altitude by N ( c m -3) = 2 0 0 e -2(1-~ for G ___0.15 = 36.5 e -15 (.15 -cy) for (y < 0.15 where •= pips, and Ps is the surface pressure. These particles simulate a possible source of marine background particles such as sea salt or marine sources of organic aerosol. The global average annual mean radiative forcing~ is listed in Table 3. The cloud forcing by carbonaceous aerosols decreased from 4.4Wm -2 to-2.4Wm -z. Figure 4 shows the temporal distribution of this forcing, with largest forcing in April and September associated with tropical biomass burning of savanna and forested areas. CONCLUSIONS Greenhouse forcing has been estimated to be about 2.5 Wm -2 over the last 100 years. Calculations shown here demonstrate that cloud forcing that results from anthropogenic emissions of carbonaceous aerosols may substantially mask the forcing by greenhouse gases. Further work is needed to appropriately estimate the pre-existing aerosol in order to better quantify this forcing. In addition, inclusion of the effect of black carbon on the single scattering albedo of clouds is needed.
Penner et al.
768 I
'
I
'
I
'
I
'
I
'
I
'
I
'
I
'
I
'
I
'
i
,
I
'
I
.
I
-2 -3 -
o
.>_ ~ - 4
~-5
ccv
-6
~'~"O I l
Jan
,
I
Feb
,
i
,
Mar ""
.............. l~.2r
A W
i
.
Apr - -
-
i
.
i
May Jun
.
i
Jul
,
i
.
Aug
i
Sep
.
Oct
Nov
i
Dec
carbonaceous aerosols carbonaceous aerosols with prescribed marine background particles
Figure 4. Cloud forcing from anthropogenic carbonaceous aerosols. ACKNOWLEDGMENTS This work was performed under the auspices of the U.S. Department of Energy under Contract W-7405Eng-48. Support from the NASA Aerosol Program and the DOE ARM Program is gratefully acknowledged. REFERENCES Atherton, C.A., J.E. Penner, J.J. Walton, D.D. Parrish (1996) Atmos. Envirron., in press. Barnard, G., and L. Kristoferson (1985) Technical Report No 4, Earthscan, 178pp. Bodhaine, B.A. (1995) J. ofGeophys. Res., 100, 8967-8975. Boucher, O. and H. Rodhe (1994) Report CM-83, Department of Meteorology, Stockholm University, 20pp. Boucher, O. and U. Lohmann(1995) Tellus, 47B, 281-300. Bower, K. N., Choularton, T. W., Latham, J., Baker, M. B., Jensen, J. and Nelson, J. (1994) J. Atmos. Sci., 51, 2722-2732. Br6mond, M.P., H. Cachier, and P. Buat-M6nard (1989) Env. Techn. Lett., 10, 339-346. Brocard, D., C. Lacaux, J.P. Lacaux, G. Kouadio, V. Yobou6, M. Assa Achy, B. Ahoua and M. Koffi, (1995) Paper presented at the Chapman Conference, Biomass Burning and Global Change, Williamsburg, March 13-17, 1995. Chuang, C.C., J.E. Penner, and L.L. Edwards (1992) J. Atmos. Sci., 49, 1264-1275. Chuang, C.C., J.E. Penner, K.E. Taylor, and J.J. Walton (1994) in Conference on Atmospheric Chemistry, January 23-28, 1994, Nashville, TN, 170-174. Chuang, C.C., and J.E. Penner (1995) Tellus, 47B, 566-577. Chuang, C.C., J.E. Penner, K.E. Taylor, A.S. Grossman, and J.J. Walton (1995) J. Geophys. Res., in press. Cooke, W.F., and J.J.N. Wilson (1995) J. of Geophys. Res., in press. Chylek, P. and J.A. Coakley (1974) Science, 183, 75-77. Crutzen, P.J. and M.O. Andreae (1990) Science 250, 1669-1678. Edwards, L.L. and J.E. Penner (1988) in Aerosols and Climate, Edited by P.V. Hobbs and M.P. McCormick, pp. 423-434, A. Deepak Publishing, Hampton, VA. Food and Agriculture, FAO Production Yearbook, vol. 45,265pp, 1991. Ghan, S.J., C.C. Chuang, and J.E. Penner (1993) Atmos. Research, 30, 197-221. Gray, H.A., G.R. Cass, J.J. Huntzicker, E.K. Heyerdahl and J.A. Rau (1984) Sci. Total Environ., 36, 17-25.
The contribution of carbonaceous aerosols to climate change
769
Guenther, A., C.N. Hewitt, D. Erickson, R. Fall, C. Geron, T. Graedel, P. Harley, L. Klinger, M. Lerdau, W.A. McKay, T. Pierce, B. Scholes, R. Steinbrecher, R. Tallamraju, J. Taylor, and P. Zimmerman (1995)J. Geophys. Res., 100, 8873-8892. Hao, W.M., M.H. Liu and P.J. Crutzen (1990) in Fire in the Tropical Biota, J.C. Goldammer ed., 440462. Hatakeyama, S., K. Izumi, T. Fukuyama, H. Akimoto, and N. Washida (1991) J. Geophys Res., 96, 947-958. Heintzenberg, J. (1989) Tellus, 41B, 149-160. Hoppel, W. A., Fitzgerald, J. W., Frick, G. M., Larson, R. E. and Mack, E. J. (1990) J. Geophys. Res. 95, 3659-3686. IPCC, Intergovernmental Panel on Climate Change (1994) Radiative Forcing of Climate 1994, Report to IPCC from the Scientific Assessment Working Group (WGI). Jones, A., D. L. Roberts and A. Slingo (1994) Nature, 370, 450-453. Langner, J. and H. Rodhe (1991) J. Atmos. Chem., 13,225-263. Leaitch, W. R., Isaac, G. A., Strapp, J. W., Banic, C. M. and Wiebe, H. A. (1992) J. Geophys. Res., 97, 2463-2474. Lee, I. Y., Hannel, G. and Pruppacher, H. R. (1980) J. Atmos. Sci. 37, 1839-1853. Lelieveld, J., and J. Heintzenberg (1992) Science, 258, 117-120. Liousse, C., H. Cachier and S.J. Jennings (1993) Atmos. Env., 27A, 1203-1211. Liousse, C., J.E. Penner, C. Chuang, J.J. Walton, H. Eddleman, and H. Cachier (1995) J. Geophys. Res., in press. Mahtab, F.U, and M.N. Islam (1984) Paper presented for Bangladesh Energy Planning Project Rural Energy Course Government of Bangladesh, Dhaka. Menaut, J.C., L. Abbadie, F. Lavenu, P. Loudjani and A. Podaire (1991) in Global Biomass Burning J. S. Levine ed., MIT Press, 133-142. Middleton, J.T., and E.F. Darley (1973) in Pollution : Engineering and Scientific Solutions, E.S. Barrekette, ed., Plenum, pp. 148-157. Nicholls, S. (1984) Q. J. R. Met. Soc. 110, 783-820. Pandis, S.N., S.E. Paulson, J.H. Seinfeld and R.C. Flagan (1991) Atmos. Env., 25A, 997-1008. Penner, J.E., H. Eddleman and T. Novakov (1993) Atmos. Env., 27A, 1277-1295. Penner, J.E., R.J. Charlson, J.M. Hales, N. Laulainen, R. Leifer, T. Novakov, J. Ogren, L.F. Radke, S.E. Schwartz, and L. Travis (1994a) Bulletin of the American Meteorological Society, 75, 375400. Penner, J.E.(1994b) in Changes in Land Use and Land Cover: A Global Perspective, edited by W.B. Meyer and B.L. Turner II, Cambridge University Press, 175-209. Penner, J.E., C.A. Atherton, and T. Graedel (1994c) in Global Atmospheric-Biospheric Chemistry, ed. R. Prinn, Plenum Publishing, N.Y., 223-248. Pruppacher, H.R~ and J.D. Klett (1978) Microphysics of Clouds and Precipitation, D. Reidel Publ. Co., Boston, 714 pp. Rogge, W.F., M.A. Mazurek, L.M. Hildemann, G.R. Cass, and B.R.T. Simoneit, B.R.T. (1993) Atmos. Environ. 27A, 1309-1330. Schnell, R.C., D.T. Kuniyuki, B.A. Bodhaine and A.D.A. Hansen (1994) Paper presented at the Fifth Conference on Carbonaceous Particles in the Atmosphere, Berkeley, August 23-26, 1994. Taylor, K.E., and J.E. Penner (1994) Nature, 369, 734-737. Zhang, S.Ho, M. Shaw, J.H. Seinfeld, R.C. Flagan (1992) J. Geophys. Res., 97, 20717-20729.
CLOUD
DROPLET
AND ITS CONNECTION
NUCLEATION
TO AEROSOL
PROPERTIES
STEPHEN E. S C H W A R T Z Environmental Chemistry Division, Brookhaven National Laboratory, Upton NY 11973 USA Abstract - Anthropogenic aerosols influence the earth's radiation balance and climate directly, by scattering shortwave (solar) radiation in cloud-free conditions and indirectly, by increasing concentrations of cloud droplets thereby enhancing cloud shortwave reflectivity. These effects are thought to be significant in the context of changes in the earth radiation budget over the industrial period, exerting a radiative forcing that is of comparable magnitude to that of increased concentrations of greenhouse gases over this period but opposite in sign. However the magnitudes of both the direct and indirect aerosol effects are quite uncertain. Much of the uncertainty of the indirect effect arises from incomplete ability to describe changes in cloud properties arising from anthropogenic aerosols. This paper examines recent studies pertaining to the influence of anthropogenic aerosols on loading and properties of aerosols affecting their cloud nucleating properties and indicative of substantial anthropogenic influence on aerosol and cloud properties over the North Atlantic. Keywords - Climate, aerosols, clouds, radiation INTRODUCTION In recent years awareness has increased of the possible climatic influence of anthropogenic aerosols though modification of the shortwave (solar) radiation budget. Two key mechanisms have been described. The first, the so called "direct effect", consists of enhancement of scattering of radiation by aerosols in clear (cloud-free) air; a portion of the scattered radiation is scattered in the upward direction, leaving the planet and thereby decreasing the radiant energy absorbed by earth-atmosphere system and exerting a cooling influence on the climate (Charlson et al., 1991, 1992). The second mechanism, the so-called "indirect effect", arises from an increase in the concentration of cloud droplets resulting from enhanced concentrations of aerosol particles that serve as nuclei upon which water vapor condenses to form cloud droplets (Cloud Condensation Nuclei, CCN). The increase in cloud droplet number concentration is postulated to increase multiple scattering within clouds, thereby increasing cloud-top albedo, a phenomenon originally suggested by Twomey (1971a; Twomey et al., 1984; Platnick and Twomey, 1994). The increase in cloud droplet concentration may also inhibit precipitation development, enhancing cloud lifetime and resulting in an increase in planetary shortwave albedo (Albrecht, 1989) and possibly also in the atmospheric absorption of longwave (thermal infrared) radiation by the resultant increased atmospheric loading of liquid water and water vapor. Estimates of the global and annual average magnitude of possible anthropogenic perturbation of the shortwave radiation budget through these mechanisms are of comparable magnitude to the perturbation in the longwave radiation budget of the troposphere due to increased concentrations of infrared active ("greenhouse") gases over the industrial period (IPCC, 1996 and earlier publications in this series; Charlson et al., 1992; Jones et al., 1994) but opposite in sign. It must be stressed, however, that these estimates are quite uncertain; indeed the uncertainty in the perturbation of the radiation budget by anthropogenic aerosols is now thought to be the greatest source of uncertainty in secular "forcing" of climate change over the industrial period. This uncertainty in turn greatly restricts the confidence that can be placed in empirical inference of climate change over this period and its qualitative and quantitative attribution to changes in radiative forcing.
770
Cloud droplet nucleation and its connection to aerosol properties
771
Much of the uncertainty in estimates of both the direct and indirect aerosol forcing arises from the present lack of ability to describe the processes governing the loading and geographical and vertical distribution of anthropogenic aerosols. This uncertainty is attributable in large part to uncertainties in the understanding of the cloud droplet nucleating properties of aerosols that govem not only the indirect effect but also the removal of aerosol particles from the atmosphere in precipitation. These uncertainties are manifest on local scales, which are suitable for study, for example, by aircraft measurements, and afortiori in extrapolation to subhemispheric scales pertinent to the transport of these aerosols within their atmospheric residence times. This paper gives an overview of the processes governing aerosol influences on climate, with particular focus on the relation between aerosol microphysical properties and the number concentration of cloud droplets. Attention is then directed toward recent studies examining this process with the aim of identifying approaches to decreasing the uncertainties in estimating the magnitude of aerosol forcing of climate on local to global scales. For recent reviews of the direct and indirect effects see Schwartz and Slingo (1996) and Schwartz (1996), respectively. DEPENDENCE OF INDIRECT FORCING ON CLOUD DROPLET CONCENTRATION The theoretical basis for a dependence of cloud albedo on cloud droplet number density, and in turn on the number density of the aerosol particles on which cloud droplets form, was given by Twomey (1974, 1977a, b), who noted the relation 10 I , ' ' ' ' '''1 between cloud albedo and optical Cloud Thickness, m depth and in turn cloud 1500 microphysical properties. This rro 0 8 d relation is exemplified in Figure 1, ~ 0 6 for which the cloud-top reflectance (albedo).RcT of a nonabsorbing, rr ,-,04 horizontally homogeneous cloud was evaluated by an analytical o 02 expression for given by Bohren o (1987) under assumption that liquid water path is constant with O0 100 1000 10 changing cloud droplet number Cloud Droplet Number Concentration, Ncd, cm" 3 concentration, Ncd. The influence Figure 1. Dependence of cloud-top albedo on cloud droplet of absorbing material in aerosol number concentration Ncd for liquid water volume fraction L = 0.3 particles on the albedo cm 3 m-3 and indicated values of cloud thickness. From Schwartz enhancement due to increasing and Slingo (1996). droplet concentrations was examined by Twomey et al. (1984) who found that except for clouds having very high albedo (RcT > 0.8), the albedo enhancement dominates over the albedo decrease due to absorbing material for reasonable assumptions of the magnitude of the absorption. o E o
_
o I-
I
i
i
i
i
i
1 i I
1
1
1
I
I
1
I
II
t
Examination of the sensitivity of cloud-top albedo to change in cloud droplet concentration yields, within the same assumptions,
ARCTIL, zc = -13[RcT(1- ~ ) ] A i n N c d .
(1)
The quantity R ~ ( 1 - RCT) exhibits maximum value of 1//4 when cloud-top albedo RCT = 1/2, for which ARcT assumes a maximum value AR~Tax = ~2 A In Ncd. As shown in Figure 2, dRcT / d In Nca varies only slowly with RCT for intermediate values of RCT, 0.3 to 0.7, characteristic of the prevalent and climatically important marine stratus clouds.
Schwartz
772 0.10 "1o E m ~3
I
I
I
I
0.4
0.6
0.8
0.08
0.06 ~>" 0.04 ._ cQ) o)
0.02
0.00
0.0
I 0.2
1 1 .0
Cloud-Top Reflectance, RCT Figure 2. Dependence of sensitivity of cloud-top albedo RCT to a logarithmic change in concentration of cloud droplet number concentration Ncd, as a function of RCT. The shaded band indicates the region, 0.28 < RCT < 0.72, for which the approximation dRcT/dlnNcd=O.075holds within 10%.
To gain a sense of the magnitude of albedo change and forcing that would be associated with a possible anthropogenic perturbation in Ncd, we note that a 30% relative increase in Ncd corresponds to an increase in absolute cloud-top albedo by about 2%. This high sensitivity to Ncd, together with the large amount of shortwave power reflected by clouds, is the basis of global estimates of the indirect aerosol forcing effect. Charlson et al. (1992; cf. also Kaufman et al., 1991) estimated the global mean
Modified from Charlson etal. (1992).
forcing due to anthropogenic aerosols in this way based on the fractional global coverage by marine stratus clouds. For a given cloud-top albedo perturbation the corresponding perturbation in top-of-atmosphere (TOA) albedo was taken as ARTOA = T 2 ~ ' T
(2)
where T is the fraction of incident shortwave radiation transmitted by the atmosphere above the cloud layer. To obtain the change in global- or hemispheric-mean albedo due only to the change in albedo of Aomst the albedo change given by (2) must be decreased further by the marine stratus clouds z.x~TOA, fractional coverage by marine stratus clouds, Amst, yielding a ,,, omst ARTOA = ~mstamTOA .
(3)
The corresponding perturbation in global- or hemispheric-mean shortwave forcing was estimated as AFC = - F ARToA = - ( F T / 4)z3d~TOA,
(4)
where F is the global- or hemispheric-mean top-of-atmosphere shortwave radiation evaluated as F = FT / 4, where FT is the solar constant. The negative sign denotes a cooling tendency. The forcing depends linearly on the perturbation in cloud-top albedo and thus exhibits a logarithmic dependence on Ncd given by (1). AFc = --0.075(FT / 4)Amstr2AlnN.
(5)
Figure 3 shows the dependence of global- or hemispheric-mean radiative forcing on a change in Ncd and indicates a sensitivity comparable to that estimated above. For an assumed 30% increase in hemisphericmean Ncd, the hemispheric-mean forcing evaluated by this approach is -1.1 W m-2; it must be emphasized, however, that the premise of the estimate (30% hemispheric enhancement in cloud droplet number concentrations) remains at present little more than an educated guess. It is clear, nonetheless, that even modest increases in the concentrations of cloud droplets by anthropogenic aerosols can lead to shortwave radiative forcing of climate that is substantial in the context of longwave forcing by anthropogenic greenhouse gases, about +2.5 W m -2 at present, global and annual average. As shown below (see also Schwartz and Slingo, 1996) the anthropogenic perturbation on aerosol mass loadings and resultant concentrations of CCN and cloud droplets can be several fold, at least on scales of 1000 km or more.
Cloud droplet nucleation and its connection to aerosol properties
773
Much work remains to be done before knowledge of the magnitude of the indirect ~. o.1~ ~ o.o2o -~ -..1 ,-, o.1o ~ - o.o6r. I ~ 6~ forcing by anthropogenic ~- 0 . 0 8 o -,-aerosols, as well as that of t--~ - 0.04"~ / ~ 4 ~9 the direct forcing, can be 0.06 -t O.OLO o ~ refined to an uncertainty 2~. 0.02 ...................................................... ~ .... d comparable to that associated =~ w i t h anthropogenic 0.00 . . . . . 0.00 0.000 1.3 2 3 4 5 6 --greenhouse gases (Penner et R e l a t i v e N u m b e r D e n s i t y of C l o u d D r o p s al., 1994). Key sources of Figure 3. Calculated perturbation in cloud-top albedo (left ordinate), uncertainty arise from the top-of-atmosphere albedo above marine stratus, global-mean albedo, and mass loading and global-mean cloud radiative forcing (right ordinates) resulting from a geographical and vertical uniform increase in cloud droplet number concentration Nccl by the distribution of anthropogenic factor indicated in the abscissa. The global-mean calculations were made with the assumption (Char]son et al., 1987) that the perturbation affects aerosols, and the only non-overlapped marine stratus and stratocumulus clouds having a microphysical properties of fractional area of 30%; the fractional atmospheric transmittance of the aerosols governing their shortwave radiation above the cloud layer was taken as 76%. The dotted cloud nucleating properties line indicates the perturbations resulting from a 30% increase in Ncd. and their atmospheric Modified from Charlson et al. (1992). residence times. As noted above, changes in cloud microphysical properties can further influence cloud short- and longwave radiative properties through changing the persistence of clouds against precipitation. However no estimates of the global forcing resulting from changes in cloud persistence appear yet to have been given. fib
0.14 -
-
0 . 0 8 1,.-,, >
-~
0.025
8~
-
0~
RELATION BETWEEN AEROSOL PROPERTIES AND CLOUD DROPLET CONCENTRATIONS Cloud droplets form on existing aerosol particles by the process of heterogeneous nucleation. Soluble material (mainly inorganic salts) provides a favorable site for nucleation by reducing the vapor pressure of the water relative to that of the pure solvent at the same particle radius that assists in overcoming the free energy barrier to cloud droplet activation. This process, initially described by K6hler, is treated in detail for example by Pruppacher and Klett (1980) and is not reviewed here; attention is called also to a recent paper (Reiss and Koper, 1995) which gives a rigorous thermodynamic treatment of the phenomenon. The thrust of such analyses is the concept of a "critical" set of values of drop radius and supersaturation, such that if the supersaturation exceeds the critical value for sufficiently long for the drop to grow to the critical radius, the droplet continues to accrete water and becomes a cloud droplet. In the ambient atmosphere the environmental supersaturation is governed by the interplay of the updraft velocity, which may be thought of as "generating" supersaturation, and condensation of water vapor on aerosol particles and growing cloud droplets, which provide a "sink" for supersaturation. This basic interplay has served as the basis for analytic and numerical examination of the dependence of cloud droplet concentration on the concentration and microphysical properties of aerosol particles and on such macrophysical environmental factors as relative humidity, lapse rate, vertical velocity, and turbulence (e.g., Roesner et al., 1990; Kaufman and Tanr6, 1994; Gillani et al., 1995; Leaitch et al., 1996). For given macrophysical properties the maximum environmental supersaturation decreases with increasing concentration of aerosol particles because of the greater surface area and sink rate for supersaturation afforded by the greater particle surface area. Nonetheless the supersaturation spectrum, the dependence of number density of aerosol particles activated to cloud droplets as a function of applied supersaturation
Schwartz
774
~0..... ' \ ................. ~; I--(NH,)~SO, I i q ~
~\ !\
i I ..... NH4HSO4I
I-
..................................
1OS
-
i l .....
! ~l
NaCl I t caso, I:'1
..............................
i i
i i
", i \i
is a key property necessary to describe cloud activation and the number density of cloud drops formed, and the dependence of this quantity on aerosol properties. For a single particle the dependence of critical supersaturation, Sc -1 (where S is the saturation ratio), on dry particle radius r0 can be evaluated from the Krhler equation under assumption that the particle consists of a single solute of molecular weight Ms, van't Hoff factor i, and density Ps (cf. Pruppacher and Klett, 1980, p. 141 ff.; also Junge and McLaren, 1971) as
Sc_l=(32~l/2(Mw)(Msll/21 \27J
~
\/~sJ
a 33/2r03/2 ,~
(6)
where cr is the surface tension and the remaining symbols have their customary meaning (the subscript w refers to Figure 4. Dependence of critical water) and in any event are defined in Pruppacher and Klett. supersaturation on dry diameter of If for an ambient aerosol the number distribution of particles aerosol particle serving as cloud as a function of dry radius is known and the composition is condensation nucleus. Particle is known or assumed, equation (6) may be used to infer the assumed to consist entirely of salt supersaturation spectrum. This dependence is shown in indicated. Figure 4, which indicates a strong dependence on particle dry diameter and a weak dependence (especially in a log-log plot) on particle composition. Equation (6) and Figure 4 provide the link between aerosol size distribution (number concentration of particles versus diameter) and supersaturation spectrum (number concentration of CCN versus supersaturation). 0.01
0.1 1 Dry diameter, IJm
10
ANTHROPOGENIC INFLUENCE ON NORTH ATLANTIC AEROSOL The North Atlantic provides a valuable testbed for studies of the influrnce of anthropogenic continental emissions on aerosol loadings and properties and on clouds because of the high spatial and temporal contrast between situations where the aerosol is predominantly anthropogenic transported from the adjacent continents versus situations of relatively pristine marine aerosol. Previous studies documenting the anthropogenic influence on the North Atlantic aerosol and its variability include Schwartz (1988) Hoppel et al. (1990), Arimoto et al. (1992), and Van Dingenen et al. (1995). Calculations of concentrations of aerosol sulfate with an Eulerian model (Benkovitz et al., 1994) indicate that much of the loading of this material over the North Atlantic results from transport of industrially emitted material from the continents. These calculations also indicate substantial spatial and temporal variability of loading of this material resulting from synoptic-scale variability in transport winds and precipitation. Figure 5 shows an example of the calculated distribution of column burden of sulfate from European anthropogenic sources over the European continent and the North Atlantic at 00 UTC on April 6, 1987. The preceding days had been characterized by a strong low pressure system that was responsible for circulating sulfate from Northern European sources well out into the North Atlantic. European emissions were the dominant source of sulfate in the indicated region during this time period. Substantial spatial and temporal variability in aerosol loadings and properties is evidenced in measurements from the ASTEX-MAGE project conducted in the vicinity of the Madeira-Azores-Canaries Islands during June 1992. Figure 6 shows a time series of sulfate concentrations at Santa Maria Island, Azores, with high excursions associated with trajectories from the British Isles and northern continental Europe (Harrison et al., 1996; Huebert et al., 1996a). Numerous measurements of aerosol loadings and size distributions taken during this project are consistent with that interpretation (Jensen et al., 1996; Russell et al., 1996).
775
C l o u d droplet nucleation and its connection to aerosol p r o p e r t i e s
Figure 5. Column burden of sulfate aerosol derived from European sources for April 6, 1987 at 00 UTC. Sulfate concentrations were calculated with an Eulerian chemical transport and transformation model that is driven by observation-derived meteorology, specifically the 6-hour forecast fields calculated by the European Center for Medium Range Weather Forecasts (ECMWF). Emissions include anthropogenic and natural SO2, non-seasalt sulfate, and reduced sulfur gases (mainly dimethylsulfide, DMS). Chemical transformation includes clear-air and in-cloud oxidation of SO2 and clear-air oxidation of DMS. Material is removed by wet and dry deposition. The model extends from -140 ~ to +62.5 ~ longitude (west of North America, across North America, the North Atlantic, and Europe, to the Urals) and has 1.125 ~ resolution. There are 15 levels in the vertical extending from the surface to about 100 hPa (the model uses ECMWF "eta" coordinates to conserve mass). Column burdens depicted here are evaluated as the vertical integral of concentration. The model is fully described in Benkovitz et al. (1994), which presents results for a onemonth simulation during fall 1986, comparisons with observations, and various statistics characterizing the model output. Figure provided by C. Benkovitz. Europeon Contribution to Sulfote Burden for Apr 6, 1987 ot OZ 8o
_- 9~ ~ o
"
. . . .
60
"
'
~ -
9- = "
~8o
:-~
#
200
"~9 4 0
,
.
"~ E zl.
20 1 - 1 O0
-50
0
50
Longitude
Examples of differences in aerosol and cloud microphysical properties in maritime and continental air are shown in Figure 7, which presents two sets of vertical profiles obtained at -1000 km separation on 16 June 1992o The concentration of accumulation mode particles (0.1 - 3 lxm diameter) is greater in continental air by a factor of ca. 30, --1500 cm -3 i , i vs. - 5 0 cm -3 and the cloud droplet concentration Santa Maria Island by a factor of ca. 5, 250 cm -3 vs. 50 cm -3. Although increase in cloud droplet content may be attributable in part to increased liquid water content, it is clear from the decrease in cloud E drop effective radius that the increase in drop 3 --i r4 t~ concentration is due mainly to the increased concentration of aerosol particles serving as t-2 cloud drop nuclei. It may be observed from + Figure 3 that such a fivefold enhancement in cloud droplet concentration gives rise to a local 1 increase in TOA albedo of--7%, corresponding, for solar irradiance of 1000 W m-2, to an instantaneous shortwave forcing o f - 7 0 W m -2. o Indeed the A V H R R visible satellite image 5 lO 15 20 Date (June, 1992) obtained at the time of the measurements shows marked increase in cloud reflectivity in the Figure 6. Time series (8-hour samples) of non seasalt continental air vs. the maritime air (Albrecht et sulfate concentration measured during the ASTEXal., 1995). MAGE project at Santa Maria Island, Azores (Harrison et al., 1996).
776
Schwartz
, 2 o o 1. . 1 , , ,......... ........ ....... ...+~!!...
+,, . . . . . . .
~..,,....,.[~..,,.,+,,
,c,f[
.... , ......
'~
Measurements of the size distribution of aerosol chemical composition are notoriously +oooi..l.... --+,: , difficult because of the small .........................."~ ................... ;i;:" .... , amount of mass contained in particles of diameter below --. "ff m~'-]:~2Z"?JJ2"':""~ .;" 0.5 ILtm, and because of the .~6oo Lill]difficulty in p h y s i c a l l y 9 s e p a r a t i n g this material. Consequently volatility measurements have long been used as a surrogate to determine the relative role of seasalt (which is non volatile) v s . 0 I,I......... I ......... I ......... I . . . . . . . . . . . . . . . 1. . . . . . . . . . . I . . . . -r:L}~-~ sulfate salts, which are volatile. 0 100 200 300 0 5 10 15 0.0 0.2 014 0 1 6 0 500 10001500 2000 Such an approach, introduced FSSP conc (/cc) FSSP EFF RADIUS (micron) J/W LWC (g/kg) PCASP conc (/cc) by Twomey (1971b), has F i g u r e 7. V e r t i c a l p r o f i l e s o f a e r o s o l and c l o u d m i c r o p h y s i c a l indicated that volatile sulfate properties over the North Atlantic in the vicinity of the Madeira salts are the dominant source of Islands on June 16, 1992, obtained with the UK Meteorological CCN in both marine and Research Flight C-130 aircraft during the ASTEX-MAGE project. continental air, a result which Solid lines represent measurements in a maritime air mass; dotted has been found to hold in the lines, continental air. a) Cloud droplet concentration; b) Cloud A S T E X - M A G E project by droplet effective radius; c) Liquid water content; d) Concentration of aerosol particles in diameter range 0.1-3.0 gm (increase indicated in Hudson and Da (1996). cloud is instrument artifact). From Albrecht et al. ( 1 9 9 5 ) . Likewise in the same project Clarke e t al. ( 1 9 9 6 ) showed that the great majority of aerosol volume in the diameter range 0.15-0.6 Ixm is volatilized at temperatures below 300~ albeit with a small residual nonvolatile core, indicative of an internally mixed aerosol consisting mainly of sulfate salts. Also in the ASTEX-MAGE project Huebert e t al. (1996) presented measurements of the mixing ratio of aerosol substances size-resolved over the diameter range 0.1 - 10 I.tm. Representative composition spectra for "clean marine air masses" and "heavily polluted continental air masses" are shown in Figure 8. In both cases the chemical amount mixing ratio of the aerosol is dominated by Na and C1, but this material, undoubtedly seasalt, is confined mainly to the large end of the size range. Ammonium and sulfate (in a near 1-to-1 ratio indicative of ammonium bisulfate) is present mainly in the accumulation mode, 0.1 to 1 I.tm diameter. Figure 9a shows the particle number concentrations for NaC1 and NH4HSO4 evaluated under the assumption that the ionic species are present in particles of those compositions. It is clear that advected continental air contains substantially more ammonium bisulfate particles in the accumulation mode than does the marine air. Despite the low molar mixing ratios for NH4 + and SO42- at the low end of the diameter range, the particle number concentrations continue to increase strongly with decreasing diameter in this region. This suggests that substantial numbers of particles are present at sizes below the range for which the chemical measurements are made, a situation that presents a very demanding challenge to measurements or, more practically, requires inferences on composition to be made by alternative means, such as volatility as noted above. As noted, measurements of the volatility of particles and CCN in this project are entirely consistent with the dominant species in this range being volatile sulfate salts, but as noted by Novakov and Penner (1993), the possible role of organics needs to be considered as well.
fl
,a,
777
Cloud droplet nucleation and its connection to aerosol properties T
a
14oo t
. . . . . . . . Marine
1200 I-
I -~
! lOOO L
.x
400
I
b
~-~
I_
s~
IL 600
Na
. . . . . . .
i + c~ I I-*- NH4+I
I
6
'
E
/!
I ~ N03I
~_
'
'
Continental
1200 -
I -e-Na
looo
d
l
1 I I
. . . . . . . .
I
"0
/
/
_
18 400 200
200
I-1
-
o
1
/
I ~ N03I I ~ MS*I
-
,,t..., Q.
'
I~c' I I + Nm+ I so, - I
-
E
i
'
14oo -
0.1
=
~ ~.~'=
~..=. ~ ~
~=
1
~ _4 ~ ~
10
[] [3
o
0.1
="
2 ~"
~t s 6=f e o I ~
3
D i a m e t e r , pm
2"
3
"4
s s /='8 10
Diameter, pm
Figure 8. Chemical composition spectra of aerosols during ASTEX-MAGE. a) Size distribution of major ionic species representative of "clean marine air masses"; b) "heavily polluted continental air masses". Data of Huebert et al (1996b). Figure 9b shows the supersaturation spectra calculated, by Eq (6), for the particle number distributions given in Figure 9a. The continental air contains much greater concentrations of CCN than does the marine air, by a factor of 2 to 6, depending on maximum supersaturation encountered. Despite the much greater mass loading of NaC1 vs. NH4HSO4, NH4HSO4 dominates the CCN for both air mass types in the supersaturation range 0.01 to 0.1%, important for marine stratus clouds. Also shown for comparison are CCN spectra for continental and marine air measured from the UK Meteorological Flight C-130 during the ASTEX project. In the limited range of overlap the agreement is quite good, both in 1
a 100
r
I
|
-..
~
i
i
,i
I
|
i
i
,
i
I
ii
,i
b
H4HSO4
1200 _
c" o
lO
c" o
=..
1
.
i
1
i
i
i
||11
,
i
~
i
1000
l-
I+.
, -o- Calculatedl -e-Measured I
-
!
a.
$
8 800
,,'
Z 0
E
r
0.1
0.01
,/,
:~ --~:
i-e",-G-Marine
I
D ,,~
-
! i
O .s
_
i.,
o
"'\.
I l l ,
i
8
9
8
I
1400-
II
,,
,,
,,
." 400
-
Co~nen
,,'"
~1~
,,,,'
ta!_..,""
"6
/
9....... 4
0.1
1
10
~
-2
3
, s eta9
0.1
I
~ I ~ 67ao
Supersaturation,% Figure 9. a) Cumulative particle number spectra (i.e., number of particles of diameter greater than value given by abscissa) for inorganic salts evaluated from measurements of Huebert et al. (1996b) shown in Figure 8 under assumption that Na and CI in the several size ranges are present as NaCI particles and likewise that NH4 + and SO42- are present as NH4HSO4 particles, b) Calculated supersaturation spectra. corresponding to these size distributions (open symbols) and supersaturation spectra representative of continental and marine air measured on the UK Meteorological Flight C-130 during the ASTEX project (filled symbols) as reported by Albrecht et al. (1995). D i a m e t e r , pm
Schwartz
778 I
I
I
I
I
,i -
, ee
_
,,' 9
~'~
80O _
z
600 -
/
,
Hercules C-130 Flights 9
---
424 + 288 [S042-]
_
,'
,
/
9
~
_
I
9
w-i
U Wash C-131A
~ o
. . . ~ . . . o .....
.................................................
..................................
94.5 + 15.5
o
0
_
[S042-1
nss [S042"1, pg m "3
Figure 10. Scatter plots of aircraft measurements of CCN concentration (1% supersaturation) vs. nonseasalt sulfate measured during the ASTEX-MAGE campaign in June 1992. Data from the UK Meteorological Flight C-130 (Harrison et al., 1996) and from the University of Washington C-131A (Hegg et al., 1993).
absolute numbers and in d e p e n d e n c e on supersaturation. A persistent issue in assessing the impact of anthropogenic aerosols on marine stratus clouds is the relation between CCN concentration and n o n - s e a s a l t sulfate, the p r e s u m e d major contributor to anthropogenic CCN. Figure 10 shows the dependence of CCN concentration on non-seasalt sulfate concentration measured in the ASTEX-MAGE project on two different aircraft, the UK Meteorological Flight C-130 and the University of Washington C- 131A. In both data sets a rather robust relation is indicated. H o w e v e r the slopes of the relations are substantially different, a situation that must be resolved.
CONCLUSIONS The magnitude of the hypothesized indirect aerosol forcing of climate by enhancing cloud albedo rests to great extent on the assumed widespread influence of anthropogenic aerosols in the marine atmosphere. Recent measurements in the North Atlantic provide c o n v i n c i n g evidence of marked enhancement (several fold) of aerosol sulfate, aerosol humber, CCN, and cloud droplet concentrations, at least at distances of a few thousand kilometers from the source of the aerosol material, albeit with considerable short range spatial and temporal variability due mainly to variability in transport meteorology. The magnitude of the enhancement in cloud droplet concentrations and inferred cloud albedo suggests substantial radiative forcing by the indirect (cloud brightening) mechanism.
Acknowledgment. This research was supported by the Environmental Sciences Division of the U.S. Department of Energy (DOE) as part of the Atmospheric Chemistry and Atmospheric Radiation Measurement Programs and was performed under the auspices of DOE under Contract No. DE-AC02-76CH00016.
REFERENCES Albrecht B. A. (1989) Aerosols, cloud microphysics, and fractional cloudiness. Science 245, 1227-1230. Albrecht B. A., Bretherton C. S., Johnson D., Schubert W. H. and Frisch A. S. The Atlantic Stratoumulus Transition Experiment--ASTEX. Bull. Amer. Meterol. Soc. 76, 889-904 (1995). Arimoto R., Duce R. A., Savoie D. L., and Prospero J. M. (1992) Trace elements in aerosol particles from Bermuda and Barbados: Concentrations, Sources and Relationships to Aerosol Sulfate.. J. Atmos. Chem. 14, 439-457. Benkovitz C. M., Berkowitz C. M., Easter R. C., Nemesure S., Wagener R. and Schwartz S. E. (1994) Sulfate over the North Atlantic and adjacent continental regions: Evaluation for October and November 1986 using a three-dimensional model driven by observaton-defived meteorology. J. Geophys. Res. 99, 20725-20756. Bohren C. F. (1987) Multiple scattering of light and some of its observable consequences. Am. J. Phys. 55, 524-533. Charlson R. J., Lovelock J. E., Andreae M. O. and Warren S. G. (1987) Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate. Nature 326, 655-661. Charlson R. J., Langner J., Rodhe H., Leovy C. B. and Warren S. G. (1991) Perturbation of the Northern Hemisphere radiative balance by backscattering from anthropogenic aerosols. Tellus 43AB, 152-163. Charlson R. J., Schwartz S. E., Hales J. M., Cess R. D., Coakley J. A. Jr., Hansen J. E. and Hofmann D. J. (1992) Climate forcing by anthropogenic aerosols. Science 255, 423-430. Clarke A. D., Porter J. N., Valero F. P. J., and Pilewskie P. (1996) Vertical profiles, aerosol microphysics, and optical closure during the Atlantic Stratocumulus Transition Experiment: Measured and modeled column optical properties. J. Geophys. Res. 101, 4443-4453. Gillani, N. V., Schwartz, S. E., Leaitch, W. R., Strapp, J. W., and Isaac G. A. (1995) Field observations in continental stratiform clouds: Partitioning of cloud particles between droplets and unactivated interstitial aerosols. J. Geophys. Res. 100, 18687-18706.
Cloud droplet nucleation and its connection to aerosol particles
779
Harrison R. M., Peak J. D. and Msibi M. I. (1996) Measurements of airborne particulate and gaseous sulphur and nitrogen species in the area of the Azores, Atlantic Ocean. Atmos. Environ. 30, 133-143. Hegg D. A., Ferek R. J. and Hobbs P. V. (1993) Light scattering and cloud condensation nucleus activity of sulfate aerosol measured over the Northeast Atlantic Ocean. J. Geophys. Res. 98, 14887-14894. Hoppel W. A., Fitzgerald J. W., Frick G. M., Larson R. E., and Mack E. J. (1990) Aerosol size distributions and optical properties found in the marine boundary layer over the Atlantic Ocean J. Geophys Res. 95, 3659-3686. Huebert B. J., Pszenny A., and Blomquist B. (1996a) The ASTEX/MAGE Experiment. J. Geophys. Res. 101, 4319-4329. Huebert B. J., Zhuang L., Howell S., Noone K. and Noone B. (1996b) Sulfate, nitrate, methanesulfonate, chloride, ammonium, and sodium measurements from ship, island and aircraft during the Atlantic Stratocumulus Transition Experiment/Marine Aerosol Gas Exchange. J. Geophys. Res. 101, 4413-4423. Hudson J. G. and Da X. (1996) Volatility and size of cloud condensation nuclei. J. Geophys. Res. 101, 4435-4442. IPCC (Intergovernmental Panel on Climate Change, 1996). Second Assessment Synthesis of Scientific-Technical Information Relevant to Interpreting Article 2 of the UN Framework Convention on Climate Change. Cambridge University Press, Cambridge. In press. Jensen T. L., Kreidenweis S. M., Kim Y., Sievering H., and Pszenny A. (1996) Aerosol distributions in the North Atlantic marine boundary layer during the Atlantic Stratocumulus Transition Experiment/Marine Aerosol Gas Exchange. J. Geophys. Res. 101, 4455-4467. Jones A., Roberts D. L. and Slingo A. (1994) A climate model study of indirect radiative forcing by anthropogenic sulphate aerosols. Nature 370, 450-453. Junge C. E. and McLaren E. (1971) Relationship of cloud nuclei spectra to aerosol size distribution and composition, J. Atmos. Sci. 28, 382-390. Kaufman Y. J. and Tanr6 D. (1994) Effect of variations in supersaturation on the formation of cloud condensation nuclei. Nature 369, 45-48. Kaufman Y. J., Fraser R. S. and Mahoney R. L. (1991) Fossil fuel and biomass burning effect on climate--heating or cooling? J. Climate 4, 578-588. Leaitch W. R., Banic C. M., Isaac G. A., Couture M. D., Liu P. S. K., Gultepe I., and Li S.-M. (1996) Physical and chemical observations in marine stratus during the 1993 NARE: Factors controlling cloud droplet number concentrations. J. Geophys. Res. In press. Novakov T. and Penner J. E. (1993) Large contribution of organic aerosols to cloud-condensation nuclei concentrations. Nature 365, 823-826. Penner J. E., Charlson R. J., Hales J. M., Laulainen N., Leifer R., Novakov T., Ogren J., Radke L. F., Schwartz S. E. and Travis L. (1994) Quantifying and minimizing uncertainty of climate forcing by anthropogenic aerosols. Bull. Amer. Meteorol. Soc. 75, 375-400. Platnick S. E. and Twomey S. (1994) Determining the susceptibility of cloud albedo to changes in droplet concentration with the advanced very high resolution radiometer. J. Appl. Meteorol. 33, 334-347. Pruppacher H. R. and Klett J. D. (1980) Microphysics of Clouds and Precipitation. (edited by D. Reidel, Hingham, MA. Reiss H. and Koper G. J. M. (1995) The Kelvin relation: Stability, fluctuation, and factors involved in measurement. J. Phys. Chem. 99, 7837-7844. Roesner S., Flossmann A. I., and Pruppacher H. R. (1990) The effect on the evolution of the drop spectrum in clouds of the preconditioning of air by successive convective elements. Quart. J. Roy. Meteorol. Soc. 116, 1389-1403. Russell L. M., Huebert B. J., Flagan R. C., and Seinfeld J H. (1996) Characterization of submicron aerosol size distribution from time resolved measurements in the Atlantic Stratocumulus Transition Experiment/Marine Aerosol Gas Exchange. J. Geophys. Res. 101, 44694478. Schwartz S. E. (1988) Are global cloud albedo and climate controlled by marine phytoplankton? Nature 336, 441-445. Schwartz S. E. and Slingo A. (1996) Enhanced shortwave cloud radiative forcing due to anthropogenic aerosols. In Clouds, Chemistry and Climate--Proceedings of NATO Advanced Research Workshop. Crutzen P. and Ramanathan V., Eds. Springer, Heidelberg, 1996. pp. 191-236. Schwartz, S. E. The Whitehouse Effect--Shortwave radiative forcing of climate by anthropogenic aerosols: An overview. J. Aerosol. Sci., in press (1996). Twomey S. (1971a) In Inadvertent Climate Modification. Report of the Study of Man's Impact on Climate (edited by Wilson C. L. and Matthews W. H.). MIT Press, Cambridge MA, p. 229. Twomey S. (1971b) The composition of cloud nuclei. J. Atmos. Sci. 28, 377-381. Twomey S. (1974) Pollution and the planetary albedo. Atmos. Environ. 8, 1251-1256. Twomey S. (1977a)Atmospheric Aerosols. Elsevier, New York. Twomey S. (1977b) The influence of pollution on the short-wave albedo of clouds, J. Atmos. Sci. 34, 1149-1152. Twomey, S., Piepgrass, M., and Wolfe, T. L. An assessment of the impact of pollution on global cloud albedo. Tellus 36B, 356-366 (1984). Van Dingenen R., Raes F., and Jensen N. R. (1995) Evidence for anthropogenic impact on number concentration and sulfate content of cloud-processed aerosol particles over the North Atlantic. J. Geophys. Res. 100, 21057-21067.
VERTICAL AND HORIZONTAL VARIABILITY OF AEROSOL SINGLE SCATTERING ALBEDO AND HEMISPHERIC BACKSCATTER FRACTION OVER THE UNITED STATES JOHN A. OGREN AND PATRICK J. SHERIDAN 1 NOAA Climate Monitoring and Diagnostics Laboratory, also 1Cooperative Institute for Research in Environmental Science at the University of Colorado, Boulder, Colorado, USA , A b s t r a c t - M e a s u r e m e n t s over the United States reveal t h a t the singlescattering albedo and hemispheric backscattering fraction show much less variability in both the vertical and horizontal dimensions t h a n the aerosol light s c a t t e r i n g and absorption coefficients from which they are derived.
K e y w o r d s - aerosol, light scattering coefficient, light absorption coefficient, single scattering albedo, backscattering, wavelength dependence. M e a s u r e m e n t s of the optical properties of submicrometer aerosol particles were m e a s u r e d from the NOAA P-3 aircraft during the s u m m e r 1995 Southern Oxidant Study. Aerosol light sc/tttering coefficient was m e a s u r e d at three wavelengths (450, 550, 700 nm) and two r a n g e s of angular integration (7-170 ~ (asp), 90-170 ~ (abs p )) with a TSI 3563 i n t e g r a t i n g nephelometer. Aerosol light absorption coefficient (%0) was m e a s u r e d at 550 nm w a v e l e n g t h with a Radiance Research PSAP continuous light absorption photometer. The sample air was heated as necessary to m a i n t a i n a relative humidity below 40%, and a multijet impactor removed particles larger than 1 ~tm aerodynamic diameter. This m e a s u r e m e n t system allows determination of several key p a r a m e t e r s needed to evaluate the radiative climate forcing by aerosols, including the aerosol single-scattering albedo (~ o = asp / (asp + Gap)), which is a measure of the relative magnitudes of s c a t t e r i n g and absorption, and the hemispheric backscatter fraction (b = abs p / asp), which describes the a n g u l a r s c a t t e r i n g properties of the particles. The majority of the flights were in the midwest and southeastern United States at altitudes below 5 km and provide a survey of the vertical and horizontal variability of the aerosols t h a t dominate the direct aerosol radiative forcing of climate. Some flights were conducted over Colorado, allowing comparison of these aerosol properties between the h u m i d E a s t and arid West. Fig. 1 shows the vertical profiles m e a s u r e d over Colorado of ~o, b, asp (denoted by Bsp in the figure legend), and the ~ g s t r 6 m exponent (c~), which is a m e a s u r e of the wavelength-dependence of aerosol light scattering. Here, the Angstr6m exponent is defined as ~ = -log[asp(~.l)/asp(~.2)] / log[~.l/~.2] , where k I and ~2 denote the m e a s u r e m e n t wavelengths of 550 and 700 nm. The data in Fig. 1 were obtained over a large a r e a of the state, and some of the variations are due to horizontal inhomogeneity. Nevertheless, the results show fairly constant values of ~o, b, and d throughout the lower troposphere. The increased variability above 5 km results from the very low (and hence imprecise) values of the primary m e a s u r e d variables, leading to large variations in p a r a m e t e r s t h a t are defined as ratios of the primary variables. The single-scattering albedo varies in the range 0.88-0.95, and the hemispheric backscattering fraction is 0.150.18.
780
Vertical and horizontal variability of aerosol single scattering albedo
781
Fig. 1" Vertical profile of aerosol properties over Colorado, 95-06-14 7000 6000~. 5000m E E 40003000-
9
200010000
i
0.0
9
l
9
9
0.5
i
l
i
~
i
l
1.0
-----single scatter albedo
,~ngstr6m exponent
~
l
i
l
9
1.5
i
2.0
2.5
back/total scatter Bsp (1E-5/m)
A similar vertical profile is seen in Fig. 2, which shows the results obtained over the s o u t h e a s t U.S., in spite of the much higher values of light scattering (note the scale change for ~sp). Once again, values at the higher altitudes are much less reliable due to the low values of the scattering and absorption coefficients. In the boundary layer, the single-scattering albedo is 0.95 and the hemispheric backscattering fraction is 0.11. These values are somewhat different from the values obtained over Colorado, suggesting systematic differences in aerosol composition and size distribution in the two regions. However, the differences may also be due to day-to-day variations in the aerosol. Fig. 3 shows the horizontal variability observed in the boundary layer on the t r a n s i t flight from Colorado to Tennessee, where the values of ~o and b are identical to the boundary layer values shown in Fig. 2. As was the case for the vertical dimension, the derived p a r a m e t e r s (~o, b, d) are relatively constant in spite of large changes in the primary m e a s u r e d variables. Finally, Fig. 4 shows the latitudinal variability t h a t was observed in the boundary layer over the midwest U.S. (Tennessee-Indiana). Slightly more variability is seen in the single-scattering albedo (0.89-0.96), but the hemispheric backscattering fraction is once again nearly constant (0.12). In all four cases, the ~ g s t r 6 m exponent stays in the range 2.0-2.5. Although i n s t r u m e n t a l noise is a limiting factor, the observed variability in may be due to variations in the aerosol size distribution, with the larger values of corresponding to cases with smaller particle sizes. Taken as a whole, the results of this study yield values for the single-scattering albedo in the range 0.88-0.96, with more variation observed fi'om day-to-day t h a n from place-toplace (horizontally or vertically). Similar conclusions can be drawn for the hemispheric
Ogren and Sheridan
782
backscattering fraction (0.11-0.18) and the AngstrSm exponent (2.0-2.5), although b in the boundary layer was always below 0.13 except for the one vertical profile over Colorado. Although it is difficult to draw general conclusions from a one-month study, the results suggest that ground-based measurements of the light scattering and absorption coefficients of submicrometer, continental particles can be used to derive values of the single-scattering albedo, hemispheric backscattering fraction, and i&mgstrSm exponent representative of the dry aerosol throughout the lower troposphere.
Fig. 2 Vertical profile of aerosol properties over the southeast U.S., 95-07-01 6000
5000
E E v
4000 3000 2000
1000 0 0.0
0.5
1.0
single scatter albedo --, ,&,ngstr6m exponent
1.5
2.0
-" back/total scatter 9 Bsp (1E-4/m)
2.5
Vertical and horizontal variability of aerosol single scattering albedo
783
Fig. 3" East-west transect, Colorado to Tennessee, 95-06-19 (open symbols denote free tropospheric measurements) 2.5 2.0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
1,5
=
o .....
i
1.00.5-
0
0.0,
A
i
~
i
i
.
A
-105
r
)
A
m
i
9
I
A
| I
I
0
&
A
I
I
A
I
*
ml
A
I
I
I
A&
-I00
-95
-90
single scatter albedo
longitude
"-- back/total s c a t t e r
,&,ngstr6m e x p o n e n t
+
I
I
-85
Bsp ( 1 E - 4 / m )
Fig. 4 North-south transect, midwest U.S., 95-07-10 2.5 2.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 9
I._I._Z. . . . . . . . . . . . . . . . . . ._
-
....... /~...j~ ..~~.~~~_.~ ..... -_- -__.. . . . . .
= ....
= .......
___
_
_
0.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
0
9
'
i
i
35
i
i
1
.
.
.
.
36
1
i
'
'
37
i
'!
38
'
'
i
'
l' !
'
39
latitude r
single scatter albedo Angstr6m exponent
-" back/total scat t er - e - Bsp ( 1 E - 4 / m )
'
'
'
40
DIRECT EFFECTS OF BLACK CARBON AND SULPHATE AEROSOLS IN THE NORTHERN HEMISPHERE.
ALF K I R K E V ~ G ,
A R N E D A H L B A C K * , and T R O N D I V E R S E N .
Department of Geophysics, University of Oslo, PO.Box 1022 - Blindern, 0315 Oslo, Norway. * Norwegian Institute for Air Research, PO.Box 100, 2007 Kjeller, Norway. Tropospheric aerosols play an important role in regulating the amount of solar radiation absorbed and reflected by the Earth-atmosphere system. Most particles, such as sulphate and chloride, mainly scatter solar radiation, whilst elemental carbon ("black" carbon, BC) in addition leads to considerable absorption. This scattering and absorption of solar radiation by the particles constitute the direct aerosol effect. Since hygroscopic aerosol particles may also serve as cloud condensation nuclei (CCN), increased amounts imply increased cloud droplet number consentrations and smaller droplets, thus indirectly affecting the optical properties of the clouds. This is the so-called indirect aerosol effect. In this work we will study possible direct effects of BC and sulphate aerosols. PARTICLE SIZES AND COMPOSITION The basis for our calculations will be an appropriate set of assumed (natural) background aerosol chemical compositions and size distributions. The set consists of two types: continental and marine background aerosols. Using available model-calculated data for the spatial and temporal distribution of BC and sulphate for a major part of the northern hemisphere (Seland and Iversen, 1996), we first estimate a modified size distribution and a new chemical composition for the particles. A minor fraction of the added anthropogenic material is assumed mixed externally, giving rise to an increased total aerosol number consentration. The ramaining (major) fraction is assumed internally mixed with the pre-existing particles. The part created by gas phase processes assumes a size distribution in accordance with diffusion/koagulation theory. The part created by liquid phase processes in clouds is evenly ditributed to all particles exceeding a certain minimum critical radius. The final distribution and composition as function of particle radius takes condensed water into account through the K6hler equation, given the relative humidity and the particle volume fraction of hygroscopic material. RADIATION SCATTERING AND ABSORPTION Tabulated values for refractive indices as function of wavelength for the relevant aerosol components will be applied in Mie calculations to estimate changes in the gross optical properties of the modified aerosol distribution relative to the pre-existing distribution. The calculations yield effective values for extinction efficiency, single scattering albedo (~) and asymmetry factor (9) as function of wavelength. Broadband weighted means (over the solar spectrum) gives final estimates of the effective wavelength independent variables. Together with the model-calculated BC and sulphate concentrations, we then have the information needed to calculate changes in optical depths (r) for a given position and time.
784
Direct effects of black carbon and sulphate aerosols in the northern hemisphere
785
RADIATIVE FORCING The aim of this work is to estimate the resulting forcing and heating/cooling rates by use of a radiative transfer model. To calculate the necessary radiation fields (in a vertically inhomogenous spherical atmosphere), we have available a discrete ordinate multistream radiative transfer model that takes into account multiple scattering (Rayleigh and Mie) and absorption, and reflection from the ground (Dahlback and Stamnes, 1991). As a natural extension of this work, the intention is eventually to implement the routines for the radiative calculations in a general circulation model, thus making it possible to estimate how the anthropogenic BC and sulphate affects the climate on the northern hemisphere due to direct effects. To minimize the computational costs, tables or/and parameterized functions must be made, relating the necessary input variables in the radiation calculations (5;, g and T) to the according physical input variables, eg. relative humidity and BC and sulphate concentrations. PRELIMINARY RESULTS At the time of writing we are still in the process of elaborating the fundamental building blocks that are to constitute the complete theoretical framework for our calculations. Figure 1, 2 and 3 show examples of single scattering albedo (5;) and asymmetry factor (g) estimates for one specific wavelength, given a set of simplifying assumptions as summarized in the figure captions. The asymmetry factor is a measure of the degree of foreward scattering in the medium, attaining a value of g = 1 for pure foreward scattering, g = - 1 for pure backward scattering and g = 0 for isotropic scattering. The single scattering albedo similarly tells us how much of the energy that is scattered, given an extinction event. 5; = 1 for pure elastic scattering (no absorption) and 5; = 0 for pure absorption. Figure 1 and 2 indicate 5; values of 0 . 8 - 0.9 for common values of RH and CBc/Cs. Seen in fight of figure 3, lower values probably can be expected when we take the whole particle size range in to account, indicating a non-negligible absorption for BC-containing aerosols. The low values of 5; and g for smaller radii is just a result of moving from the Mie into the Rayleigh regime for scattering. The gross optical properties of a size distribution is obtained by integrating over all particle sizes and radiative wavelengths. An example of background size distribution and its modified counterpart is shown in figure 4. The method of modifying the size distribution is comparable to that of Chuang and Penner (1995). In addition to calculations of & and g, estimates of changes in CCN can be inferred from these changes in size distributions. REFERENCES Chuang, C.C., and J.E. Penner (1995) Effects of anthropogenic sulfate on cloud drop nucleation and optical properties. Tellus, 47B, 566-577. Dahlback, A., and K. Stamnes (1991) A new spherical model for computing the radiation field available for photolysis and heating at twilight. Planet. Space. Sci., 39,671-683 Seland, ~., and T. Iversen (1996) Model calculations of hemispheric scale dispersion of black carbon and sulphate. (this issue).
Kirkevgtg et al.
786 1.00
/
0.80 / i.
/
/
/
/
j
I.
0.60
0.40
single scattering albedo asymmetry factor g
0.20
0.00 0.00
.
.
0.20
.
.
.
6 O. 0
0.40
'
8 O. 0
'
1.00
relative humidity R H
Figure 1: Single scattering albedo ~ and asymmetry factor g as function of relative humidity RH, for the wavelength A = 0.55 #m. All particles are assumed to have the same dry radius r0 = 0.15 #m. The chemical composition for the dry aerosol is constituted by BC and ammonium sulphate internally mixed. The ratio of the mass concentrations for BO and sulphate used is CBc/Cs = 0.15. Note that RH = 0.6 is the assumed treshold for hygroscopicity.
1.00
0.80
0.60
0.40
0.20
0.00 0.0
single scattering albedo asymmetry factor g
'
'
0.2
'
'
0.4
'
'
0.6
'
[
08
'
~b
.0
CBc/Cs Figure 2: As in figure 1, except that if; and g are plotted as function of mass concentration ratio CBc/Cs. Particle dry radius r0 = 0.15 #m and relative humidity RH = 0.6 (barely above hygroscopicity treshold).
Direct effects of black carbon and sulphate aerosols in the northern hemisphere 1.0
,
9 .
,
,
.
787
,
/"- ~--,~ I/I/
0.8
0.6
Iiiiii
0.4
/
0.2
/
single s c a t t e r i n g
/
asymmetry
/ / I j
0.0
/
.5
albedo
factor g
/
-1.0
-0.5
0.0
0.5
1.0
log (re/tam) Figure 3: As in figure 1, except that r and g are plotted as function of particle dry radius r0. Mass concentration ratio CBc/Cs = 0.15 and relative humidity RH = 0.6.
4.0 pre-existing marine aerosols anthropogenic sulphate-containing aerosols
3.0 "x3
2.0
1.0
0.0
-3.0
.
.
.
.
-2.0
.
.
.
. .
log
.
.
-1.0
.
(d/tam)
.
.
t
0.0
1.0
Figure 4: Background number distribution dN/d (log d) (d is particle diameter) for marine aerosols with a total number concentration Ntot = 250 cm -s, and a simulated modified distribution. We have assumed that 0.5 l~g/rn3 sulphate is distributed to the pre-existing aerosols without generation of new particles.
EVALUATION OF THE "TWOMEY EFFECT" WITH NUMERICAL MODELS AND COMPARISONS WITH OBSERVATIONS"
CLOUD
A. S. ACKERMAN and O. B. TOON NASA Ames Research Center, Moffett Field, California, USA Abstract - Assessments of the effects of increased aerosol concentrations on cloud albedo, and hence on the global radiative heat budget, are highly uncertain. Estimates and measurements of the sensitivity of cloud albedo to local changes in droplet concentrations indicate that the greatest absolute changes in albedo are found where concentrations of cloud droplets (and aerosols) are lowest. However, determination of effects of pollution sources on the albedo of clouds downwind requires an evaluation of changes in aerosol concentrations during transport from their source regions. Here we use results from a numerical model to calculate the sensitivity of cloud albedo to changes in aerosol concentrations from an upwind source of pollution. We find the greatest increases in cloud albedo to occur close to pollution sources, rather than well downwind in comparatively clean regions. Keywords - Indirect effect of aerosols on climate; Cloud albedo susceptibility Aerosols can affect cloud albedo by increasing cloud droplet concentrations; the impact on the global radiative budget is referred to as the indirect effect of aerosols on climate. However, there are large uncertainties in current estimates of the magnitude of the indirect effect. Also, since tropospheric aerosols are not mixed uniformly over the globe, their radiative effects vary regionally. Twomey (1991) defined cloud albedo susceptibility (dA/dN) as the sensitivity of cloud albedo (A) to absolute changes in droplet concentration (N). Defined in this way, the susceptibility is greatest in regions where N is low, and changes in cloud aibedo due to aerosol pollution are expected to be greatest in clean air (e.g., over the remote ocean). This deduction is correct for the sensitivity of cloud albedo to local aerosol sources (such as the effects of ships on local marine clouds). However, the sensitivity of cloud albedo to a distant aerosol source is also affected by changes in the concentrations of the emitted aerosols with downwind distance from the source. The sensitivity of cloud albedo near a source of aerosols (A s) to the aerosol concentration near thesource (Ns) is given by the (local) susceptibility, dA s/dN s. (We will assume initially that changes in droplet concentrations are equivalent to changes in aerosol concentrations, although we will subsequently relax this assumption.) The sensitivity of cloud albedo well downwind of the source of aerosols (Ad) to the aerosol concentration near the source is given by dAd/dNs = (dAd/dNd)(dNd/dNs), where N d is the aerosol concentration downwind. Thus, evaluation of dAd/dNs requires a relationship between N d and N s. For purposes of illustration we assume that this relationship is of the form N d = Nd0 + czNs, where Nd0 is the ambient concentration of aerosols at the downwind location, maintained by local sources, and cz is a factor that accounts for depletion of aerosol concentrations during their transport downwind. These relations lead to
dAd / Nd~ dAd d(~ ~/s ) = 1 - N d ) d(~
~/d)
(1)
In this case, the sensitivity of cloud albedo to a source of pollution upwind (dAd/d[ln Ns]) is proportional to the susceptibility of cloud albedo to relative changes in the local aerosol concentration (dA d/d[ln Nd] = dAd/[dN a/Nd], which we will refer to as the relative susceptibility and which is always locally defined), multiplied by a factor (1 - Nd0/Nd) that decreases as polluted air is
788
Evaluation of the "Twomey effect" with numerical cloud models
789
transported downwind. If the relative susceptibility does not change downwind of a source of aerosols, the effect on the albedo of clouds downwind will generally be reduced compared to their effect near the source. The simplest estimate of relative susceptibility is made by assuming that cloud water is constant, in which case dA/d(In N) = A (l-A)/3. According to this relationship, the relative susceptibility has a broad maximum at A = 1/2, as noted by others (Twomey, 1991; Platnick and Twomey, 1994). Radke et al. (1989) measured an enhancement of cloud water in ship tracks. Platnick and Twomey (1994) parameterized the relationship between cloud water and droplet concentration and found that cloud susceptibility increased by a factor of 1.8 over the assumption of constant cloud water. Pincus and Baker (1994) used a mixed-layer model to investigate some interactions between cloud microphysics and the energetics of the boundary layer. In their results (shown in Fig. 1) cloud water increased with droplet concentrations, resulting in cloud susceptibilities that were up to 2.5 times greater than under the assumption of fixed cloud water. The increased susceptibility in their steady-state model was driven by a decrease in the precipitation reaching the surface as droplet concentrations increased. We have used a more detailed numerical model of the stratiform-cloud topped marine boundary layer to evaluate the relative susceptibility of cloud albedo (Ackerman et al., 1995a). To evaluate relative susceptibility, we ran our model with fixed boundary conditions until a steady state was reached (allowing for the variations of diurnal oscillations). The only change prescribed between model runs was the production rate of aerosols. A comparison of our model results with previous calculations of relative susceptibility is shown in Fig. la. Consistent with the analysis of Platnick and Twomey (1994) and the model results of Pincus and Baker (1994), we find that susceptibility is greater than under the assumption of fixed cloud water. However, the details of the dependence of susceptibility on droplet concentration in our results differ from previous estimates. The cloud liquid water path (at noon) predicted by our model increases with droplet concentrations (N) for N < 500 cm -3. At the lowest droplet concentrations this increase is due to a positive feedback loop between increases in cloud water, longwave radiative cooling, and turbulent mixing (Ackerman et al., 1993). At droplet concentrations between --10 and 500 cm -3, a decrease in precipitation with decreasing droplet size allows the cloud liquid water path to increase with N. However, at higher droplet concentrations (N > 500 cm-3), the cloud liquid water path during the late morning and afternoon decreases with increasing N in our model, due to the more rapid evaporation of smaller cloud droplets below cloud base. The increased evaporation reinforces the daytime reduction in boundary-layer mixing due to solar heating in the cloud layer, which decreases the supply of water vapor to the cloud layer (Ackerman et al., 1995b). In our model, the cloud liquid water path during the evening and early morning always increases with droplet concentration. Although cloud water (at noon) decreases with droplet concentration for N > 500 cm -3, the relative susceptibility continues to increase with increasing N in our model results, due to a disproportionate increase in the concentration of interstitial haze particles. This increase results from a reduction in the peak supersaturation in cloud with increasing N. Although the concentration of interstitial haze particles may be exaggerated in our model results (Ackerman et al., 1995a), an increase in the ratio of interstitial particles to cloud droplets with N increasing is qualitatively realistic,and is supported by measurements and parcel updraft models (Leaitch et al., 1986; Martin et al., 1994; Jensen and Charlson, 1984). The dramatic upturn in the relative susceptibility at high N in our model results is due to the increasing concentrations of interstitial particles that do not contribute to N. As shown in Fig. lb, counting aerosol particles (active as cloud condensation nuclei at 1% supersaturation) in addition to cloud droplets in the denominator of dA/d(ln N) significantly reduces the relative susceptibility at high particle concentrations. In contrast to the strong dependence of susceptibility on droplet
790
Ackerman and Toon
concentration in our model results (Fig. la), the overall dependence of the relative susceptibility to the total particle concentration is rather weak (Fig. 1b). Although the details of the underlying reasons are quite different, the values of our relative susceptibility to total particle concentration are surprisingly similar to Twomey's, provided that droplet concentrations are (unrealistically) equated to total aerosol concentrations in Twomey's relationship. As with Twomey's relationship, our model results predict that the relative susceptibility is generally insensitive to local aerosol concentrations (as implied by the lack of a strong overall trend in Fig. lb). However, we have only investigated the susceptibility for one set of (fixed) boundary conditions (such as sea-surface temperature and minimum solar zenith angle). Also, the implications of our results for continental clouds are uncertain. For instance, a higher surface albedo over land decreases the susceptibility of cloud albedo, and the interactions between cloud microphysics and vertical mixing may be different over the continents than over the oceans. Equation (1) shows that a constant relative susceptibility implies that the effect of aerosols on the albedo of clouds downwind will be reduced relative to their effect near the aerosol source. Hence, our model results indicate that the effects of aerosols on cloud albedo are reduced with increasing distance downwind of a pollution source, as suggested by Kaufman et al, (1991). This conclusion is consistent with analyses of satellite data indicating that the effects of aerosol pollution on the albedo of marine boundary-layer clouds are greatest near coasts downwind of pollution sources (Falkowski et al., 1992; Kim and Cess, 1993). Results from global models also support this conclusion (Jones et al., 1994; Erickson et al., 1995). Much of the research investigating the effects of aerosol pollution on cloud albedo has focused on clouds in the clean marine atmosphere; our results suggest that more attention should be focused on clouds near and over the continents (Kaufman and Nakajima, 1993). Regions near pollution sources do not necessarily exert the greatest indirect effect on the global radiative heat budget, because the effects must be integrated over area: it may be that more extensive regions of clouds less affected by aerosol pollution have a greater impact on global climate than do less extensive regions of clouds affected more strongly by pollution. Global models are appropriate tools for investigating such issues. However, when treating the indirect effect of aerosols on climate, we recommend that global models use parameterizations that are consistent with results from detailed studies such as that presented here. Otherwise one risks making incorrect assumptions, such as absolute susceptibility being independent of droplet concentrations (Erickson et al., 1995), which drastically skews regional variations in radiative forcing. REFERENCES Ackerman, A.S., Toon, O.B. and Hobbs (1993), P. V. Science 262, 226-229. Ackerman, A.S., Toon, O.B. and Hobbs (1995a), P. V. J. atmos. Sci. 52, 1204-1236. Ackerman, A.S., Toon, O.B. and Hobbs (1995b), P. V. J. geophys. Res. 100, 7121-7133. Erickson, D.J. III, Ogelsby, R.J. and Marshall (1995), S. Geophys. res. Lett. 22, 2017-2020. Falkowski et al (1992). Science 256, 1311-1313. Jensen, J.B. and Charlson, R.J. (1984) Tellus 36B, 367-375. Jones, A., Roberts, D.L. and Slingo, A. (1994) Nature. 370, 450-453. Kaufman, Y.J., Fraser, R.S. and Mahoney, R.L (1991). J. Clim. 4, 578-588. Kaufman, Y.J. and Nakajima, T. J. (1993) appl. Met. 32, 729-744. Kim, Y. and Cess, R.D.J. geophys. Res (1993). 98, 14883-14885. Leaitch, W.R., Strapp, J.W. and Isaac, G.A. (1986) Tellus 38B, 328-344. Martin, G.M., Johnson, D.W. and Spice, A. (1994) J. atmos. Sci. 51, 1823-1842. Platnick, S. and Twomey, S. J. (1994) appl. Met. 33, 334-347. Pincus, R. and Baker, M. (1994) Nature 372, 250-252. Radke, L.F., Coakley, J.A., Jr. and King, M.D. (1989) Science 246, 1146-1149. Twomey, S. (1991) Atmos. Envir. 25A, 2435-2442.
Evaluation of the "Twomey effect" with numerical cloud models
0.3
T
,T~,ww~]
'
'
'''"~l
'
~ ''''"l
0.3
a
Present Results Pincus & Baker b (1994) ...'-..
0.2
........
I
........
0.2
I
,, ~ ~ 9
........
791 I
'
' ' ~
~ ~
~
,,
O
era 1.1 ~m radius, and A is the broadband solar albedo at a height > 100 m above cloud top). For the simulations presented here the boundary conditions are representative of a summertime marine boundary layer in middle latitudes. The dotted line is from the steady-state model of Pincus and Baker, in which cloud water was allowed to vary. The dashed line is Twomey's relationship, which was derived under the assumption of constant cloud water, using the dependence of cloud albedo on N from the present model results. (b) Relative susceptibility of cloud albedo to changes in Nto~, the total number concentration of haze particles (active as cloud condensation nuclei at 1% supersaturation) plus cloud droplets. The solid line shows the results from the present model, in which Nto~ is averaged over the depth of the boundary layer. For the results of Pincus and Baker (dotted line) and Twomey (dashed line) it is assumed that Nto~ = N.
NUCLEATION AND GROWTH INTERACTION AND ITS EFFECTS ON CLIMATE
IN CLOUDS
YOU-SUO ZOU AND NORIHIKO FUKUTA Department of Meteorology, University of Utah, Salt Lake City, Utah 84112, U.S.A. Abstract - Using a newly developed cloud model with detailed microphysics, the interaction of nucleation and droplet growth at the cloud base has been studied. The maximum supersaturations generated by the model have been used to evaluate the accuracy of analytic formulas developed by Twomey (1959) and Fukuta and Xu (1996). A correction factor for the Fukuta-Xu formula, as well as the effect of condensation coefficient on cloud albedo change, are obtained. Keywords - Maximum supersaturation; Droplet number concentration; Cloud albedo change INTRODUCTION In order to find the relationship between the characteristics of cloud forming air mass and the resultant cloud albedo variation through the cloud formation process, a cloud model with detailed microphysical processes has been formulated. The process involves the nucleation and droplet growth interaction problems above the cloud base to evaluate the maximum supersaturation. The model assesses correction factors for analytic formulas developed by Twomey (1959) and Fukuta and Xu (1996). Thus, the modified Fukuta-Xu analytic formula is now allowed for use in cloud research, particularly for maximum supersaturation, droplet nucleation and number concentration as well as subsequent cloud properties for climatic balance such as albedo change under the influence of co, [3, C and k, the thermal accommodation and condensation coefficients and constants describing Cloud Condensation Nuclei (CCN) activities, respectively. Some of the research results are presented in this paper. A BRJEF DESCRIPTION OF THE NEW CLOUD MODEL The cloud model with detailed microphysics and chemistry is designed to describe two phases of convective cloud formation: the haze phase before the droplet nucleation mostly below the cloud base and the nucleation-growth phase above the cloud base. In the haze growth phase, 100 groups of dry particles of ammonium sulfate with radii ranging from 2.5x10 -3 to 1.1 lum begin to grow at initial condition: the pressure P0- 809.21 mb, the temperature To = 19.3~ relative humidity, RH - 95%. Haze particle growth is described by the modified Fukuta-Walter (1970) droplet growth equation (Fukuta and Xu, 1996). Excessive growth of these haze particles is limited by the equilibrium droplet size of the K6hler equation. The cloud droplet begins to form by nucleating haze particles from the larger size as RH exceeds 100%. The nucleation phase above the cloud base begins following the haze process at the cloud base pressure Pb = 800 mb, the cloud base temperature Tb = 10~ This is a closed, adiabatic droplet growth process, in which modified difffusion-kinetic and Maxwellian droplet growth equations are used to evaluate analytic equations of Twomey (1959) and Fukuta and Xu (1996) for maximum supersaturation the cloud parcel achieves.
792
Nucleation and growth interaction in clouds and its effects on climate
793
The droplet nucleation is controlled by the modified Kohler nucleation theory (Young and Warren, 1992). The total number concentration of activated droplets is given as N - C(S - l ) k....
(1)
where C = 100 cm -3, k = 0.5 for a maritime CCN; C = 500 cm -3, k = 0.7 for continental CCN. The cloud albedo related process and treatment in the model are taken from Fukuta et al. (1995) RESULTS AND ANALYSIS Twomey's formula (1959) for predicting the maximum supersaturation:
(S
-
l)max,T
=
i
1.6x 10 -3
}!
kB(--3/~l(i2)
3
2
_,
Zk*4 " Wu
9C
(2)
,
where B(3/2,k/2) is the complete beta function, wu the updraft velocity and C and k are parameters of the CCN spectrum (1). Fukuta et al. (1995) and Fukuta and Xu (1996) also worked out a Maxwellian formula for the maximum supersaturation: 3
(S
-
G)
1)max,F =
1
3
2~k*2)[(2k+ 3)F]Xk'~ .
-1
_ 2(k*2) Wu
9C
k +2
(3)
For A, B, F, G and other parameters in Eqs. (3), see Fukuta et al. (1995) and Fukuta and Xu (1996). 8.0
!
20.0
!
9
present model ......
6.0
Twomey Fukuta-Xu
tO
~
........... ~ .
o~
!
! ~
present model ...... Fukuta-Xu
15.0
~
~176 ~ .
~
C 0
~
.
3
4.o
10.0
ill
ffl
09
,
(1) o
2.0
0.0
09
0
5
10
15
Updraft Velocity (m/s) Fig. 1 Maximum supersaturations predicted by the present cloud model and Twomey and Fukuta-Xu formulas using Maxwellian droplet growth theory.
5.0
0.0
,
0
,
I
,
,
,
,
1
.
5 10 Updraft Velocity (m/s)
.
.
15
Fig. 2 Maximum supersaturations predicted by Fukuta-Xu formula and the present cloud model using diffusion-kinetic droplet growth theory.
Figure 1 compares the maximum supersaturations of continental cloud predicted by the new cloud model and the Twomey and Fukuta-Xu analytic formulas, using the Maxwellian droplet growth equation 9 Figure 2 is also for continental clouds but uses the difffusion-kinetic droplet growth equation, with a = 1 and [3 = 0.03. Since the Twomey formula is unable to deal realistically with the diffusion-kinetic process, results from the new cloud model and the Fukuta-Xu formula are compared in Fig. 2. Here we just present plots and examples for continental clouds. Situations for maritime clouds are also evaluated and have shown the same tendency as continental clouds 9 Details can be found in Zou (1996). Figures 1 and 2 demonstrate that, compared to the model results, both the Twomey and Fukuta-Xu formulas show
Zou and Fukuta
794
obvious errors, although the Twomey error is smaller than that of Fukuta-Xu. Therefore, a correction factor, J, is necessary to make these formulas, particularly the Fukuta-Xu formula, practical and accurate to express the effects of t~ and 13. Computed results show that J is a function of a, 13, C and k but nearly independent of w, ( S - 1)max,md,DK J(~, ~,C,k) = , (4) (S - 1 )maK,F, MXW where (S - 1)~v,~tx-w or Eq. (3) refers to the maximum supersaturation of the Fukuta-Xu formula using the Maxwellian theory. (S - 1),~,~~ DKrefers to the maximum supersaturation of the cloud model using the difffusion-kinetic droplet growth theory.
1.5,
\ll\'
ooo~
\1\
-%~oo \ fo.
9)
o
~ ^~ 0
~
0o4
u.u,-c o n d e n s a t i o n
o.o6 coefficient
o.08
o.~
j8
Fig. 3 Correction factor Jl(13,C) for maritime clouds for ct = 1 and k = 0.5. 1.4,
\
\
\\
oo
%~176 -~, 5 0 0 ~ 0 6 "0.08 0 Condensation Coe~{icient/3
0.1
Fig. 4 Correction factor J2(13,C) for continental clouds for t~ = 1, k = 0.7. Figures 3 and 4 demonstrate the configuration of J~(13,C) -- J(1, [3, C, 0.5) for maritime clouds and J2([3,C) - J(1, 13, C, 0.7) for continental clouds. These two figures show that both Ji and J2 vary with 13 and C values. For the most practically uses, J~(1, 0.03, 100, 0.5) = 0.881 applies for maritime clouds and J2(1, 0.03, 500, 0.7)= 0.9t6 for continental clouds. Once the J(ct,13,C,k) is obtained, the Fukuta-Xu analytic formula can be written as
Nucleation and growth interaction in clouds and its effects on climate Table 1
795
Cloud albedo as a function of condensation coefficient ~3.
For maritime clouds, N - 100(S - 1 ) ~
For continental clouds, N = 500(S - 1),~,
i3
Ro
R
AR/Ro(%)
13
Ro
R
AR/Ro(~)
0.10
0.5
0.553
10.6
0.10
0.5
0.597
19.4
0.03
0.5
0.557
11.5
0.03
0.5
0.603
20.6
0.01
0.5
0.566
13.3
0.01
0.5
0.616
23.1
, ,
,,
(S - 1)m~,` = J ' ( S -
1)max,F,MXW
(5)
Eq. (5) is called the Modified Fukuta Formula for Maximum Supersaturation Prediction. Using the Modified Fukuta Formula in Eq. (5) and J~ = 0.881 and J2 = 0.916 as well as albedo change equations (Twomey, 1991" Fukuta et al., 1995), we have calculated the albedo change, AR/Ro, with [3 values as shown in Table 1. In the table, Ro is the average or background albedo, R the actual albedo calculated using specific [3 values, and AR/Ro = (R - Ro)/Ro the relative albedo change due to [3 values. The table confirms that the reduction of 13values can significantly increase the cloud albedo. If the condensation coefficient [3 is reduced by anthropogenic activities, then the albedo change becomes very significant. Comparing the data for maritime and continental clouds, we can see that the albedo change for continental cloud is significantly larger than that for maritime cloud when values of 13 and Ro are the same, which is opposite to the effect of anthropogenic CCN injection. This demonstrates that the larger CCN concentration caused by air pollution will also enhance the condensation coefficient effect on cloud albedo increase. CONCLUSION A central problem for the linkage of CCN change, droplet generation and cloud albedo variation is the evaluation of maximum supersaturation in clouds. Twomey's formula, though less erroneous than that of Fukuta-Xu, is unable to deal with the diffusion-kinetic process, and the beta function involved in the former also causes some inconvenience in practical use. Fukuta-Xu analytic Maxwellian formula, which includes parameters of cloud base condition, is simple and convenient in practical use although contains larger errors as judged by the model's output. Therefore, a correction factor of the diffusion-kinetic effect has been worked out for the latter in comparison with the model output. The cloud model generates maximum supersaturation change, cloud droplet concentration and cloud albedo variation and demonstrates the effect of condensation coefficient on droplet number concentration and cloud albedo change caused by air pollution. Acknowledgment. This study was supported by the Division of Atmospheric Sciences, National Science Foundation, under Grant ATM-9112888. REFERENCES Fukuta, N. and Walter, L. (1970) J. Atmos. Sci. 27, 1160-1172. Fukuta, N. and Xu, N. (1996)Atmos. Res. (in press). Fukuta, N., Xu, N. and Zou, Y.S. (1995)Preprints, Conf. on Cloud Phys., Amer. Meteor. Soc., Jan. 15-20, 1995, Dallas Texas, 478-492. Twomey, S. (1959) Geofis. Pura. Appl. 43, 243-249. Twomey, S. (1991) Atmos. Env. 25A, 2435-2442. Young, K.C. and Warren, A.J. (1992) J. Atmos. Sci. 49, 1138-1143. Zou, Y.-S. (1996) Model Simulation of Convective Cloud Formation and Its h~uence on Cloud Optical Properties, Ph.D. Dissertation, Department of Meteorology, Univ. of Utah.
I M P A C T OF S U L F A T E A E R O S O L S ON S H O R T - W A V E A L B E D O OF C L O U D S : A FIELD S T U D Y V.K. SAXENA and S. MENON North Carolina State University, Department of Marine, Earth and Atmospheric Sciences, Raleigh, NC 27695, USA.
ABSTRACT On a regional scale, the indirect effect of aerosols is investigated by analyzing the impact of anthropogenic sulfur emissions on the microstructure and short-wave albedo of clouds formed at a mountain-top location and simultaneously observing these clouds by AVHRR (Advanced Very High Resolution Radiometer) aboard the NOAA (National Oceanographic and Atmospheric Administration) satellites. The aerosol content of the cloud forming air mass is analyzed for size and morphology by electron microscopy and x-ray energy spectrometer. The ionic composition of cloud water is measured along with meteorological parameters. The thickness of the overlying clouds and the 48 h backtrajectories of air masses are determined from on site measurements and Hybrid Single Particle Lagarangian Integrated Trajectory (HY-SPLIT) model. The short-wave albedo of clouds determined from in situ measurements decreased with an increase in cloud water pH which was largely affected by the sulfate concentration in the cloud forming air mass. For six coincidental cloud events, the AVHRR retrieved cloud albedo (channel 1, )~ = 0.63 gm) agreed well with those determined from in situ microphysical measurements, the correlation coefficient being 0.95. These findings indicate that the sulfate contents of air masses are capable of influencing the shortwave albedo of thin, continental clouds. INTRODUCTION There has been considerable theoretical discussion on the cooling effect of aerosols (e.g. Charlson et al., 1992; Ghan et al., 1990; Wigley, 1991), but the experimental evidence to validate this effect has been sketchy so far. The indirect radiative forcing mechanism due to aerosols can alone counteract the greenhouse warming effect due to the doubling of CO2 (IPCC, 1994). The role of sulfate aerosols in modifying the shortwave albedo of clouds and thereby, global climate, has been recently debated. The cloud condensation nucleus (CCN) activation spectrum as well as cloud microphysics are modified by the hygroscopicity of sulfate aerosols. The purpose of this paper is to examine the aerosol content of the different cloud forming air masses that arrive at our experimental site and help establish a relationship between the aerosol content of the air mass and the resulting cloud microphysical and optical properties. METHODOLOGY Mountain top locations offer unusual and unique opportunities for such investigations because they are often immersed in thin clouds (see for example, Saxena et al., 1989). In this paper we use results from data obtained during a field experiment conducted in Mount Mitchell (2038 m msl; 35044'05" N, 82~ W) State Park during 1993-1995. A complete description of the site and the instrumentation is given by Saxena et al., (1996). The source of the cloud forming air mass was inferred from backtrajectory analysis using a Hybrid Single Particle Lagrangian Integrated Trajectory (HY-SPLIT) model (Draxler, 1992). The optical depth (~) of the cloud is calculated as shown by Twomey (1977),
796
Impact of sulfate aerosols on short-wave albedo of clouds: A field study
797
l
( 1)
r = H { (9 / 2 ) ; r ( L W C ) 2 C D N C ( p -2 ) } 3
where the density (p) of liquid water is 1000 kg m 3 and the cloud thickness H is obtained from the HY-SPLIT sounding. LWC, the cloud liquid water content and CDNC (cloud droplet number concentration) are obtained from the FSSP (Forward Scattering Spectrometer Probe) size distributions. The albedo (A) is then evaluated using the relationship (Lacis and Hansen, 1974), A-
r
(2)
r+7.7 Equations (1) and (2) were used to calculate the in situ cloud albedo. These were compared to the ones retrieved from the Advanced Very High Resolution Radiometer (AVHRR) aboard NOAA -11 and NOAA-12 polar orbiting satellites which provide the visible (Channel 1, 1 = 0.63 mm), infrared (Channel 3, 1 - 3.7 ram), and thermal infrared (Channel 4, I = 11 mm) information. RESULTS AND DISCUSSION The site location and our criterion for designating air masses as polluted (P), continental (Co), and marine (M) based on the emission inventories of the U.S. Environmental Protection Agency are shown in Fig. 1.
Figure 1: The 1991 U.S. EPA Emissions Inventory of anthropogenic sulfur oxides (SOx) for the eastern United States.
Ambient aerosol samples were collected during the summer of 1994 using an Anderson dichotomous sampler before and after non-precipitating low-level cloud events. Table 1 contains six of these cases for which the cloud water was collected. Table 1. Summary of aerosol sampling periods, sulfate mass, back trajectories of ensuing cloud events. Cloud event
pH
29-May-94 A 29-May-94 B 23-Jun-94 20-Jul-94 9-Aug-94 A 9-Aug-94 B
3.38 _+0.06 3.53 + 0.10 2.61 + 0.28 3.68 + 0.06 3.26 + 0.26 3.31 +0.03
[SO4= ] mass Before 4.66 + 0.07 5.96 + 0.09 20.38 + 0.31 1.82 + 0.04 14.59 + 0.22 15.16 +0.23
in (gg m "3) After 5.96 + 0.09 6.01 + 0.09 17.53 + 0.26 4.06 + 0.08 15.16 + 0.23 10.25 +0.16
Source Before After PCo CoM CoM CoM C MC M M MCo P P CoP
Scanning electron microscopy was used to examine aerosol samples collected before and after cloud events. The particles in the samples are identified by their morphology and elemental
Saxena and Menon
798
constituents as mineral, organic or industrial. Back-trajectory analysis of these samples indicate the type of air masses associated with the aerosol samples. Data from the table indicate that higher sulfate masses are associated with lower pH values and vice-versa. Highest sulfate mass concentrations in fine ambient aerosol occurred for continental air masses as opposed to polluted air masses whereas lowest values occurred for marine air masses. The presence of higher sulfate masses in continental air masses could be due to the production of CaSO4 particles by the reaction of calcite with sulfuric acid in the cloud droplets (Deininger and Saxena, 1996). From the microscopic analysis of the aerosols collected, we were able to quantify the sulfate content in the aerosol sample and relate it to the source of the air mass. Further details regarding these cases can be obtained in Deininger and Saxena (1996). In order to investigate the indirect radiative forcing role of sulfate aerosols, the relationship between the sulfate content of the cloud water and the CCN concentration in the pre-cloud air needs to be examined. For the 1995 field season, the CCN activation spectra were measured by a Horizontal Thermal Gradient Cloud Condensation Nucleus Spectrometer described by Fukuta and Saxena (1979a, b). Activation spectra obtained an hour prior to the cloud event were assumed to represent the cloud forming air mass. The CCN concentration activation spectrum is represented in terms of a power function as; N = C (S) k (3) where N is the number of CCN that would be active at or below the supersaturation S (in %), and C and k are concentration and slope parameters which are obtained empirically. Good correlations were obtained when the CCN concentration at 1% supersaturation as well as the cloud albedo calculated from in situ measurements were plotted as functions of the cloud water sulfate content as can be noticed in Fig. 2. 2500
.........................
9
..................
y = 0.0089x + 40.647
~
R 2 = 0.6234
~
2000
9
*
A
-
70
-
60
9
5O ~ 104 cm 3 for Aitken nuclei, and < 1 cm -3 for particles d > 10 lam. In this duststorm the small particles decreased ten-fold and giant particles increased by two orders of magnitude. The strong winds and dense dust might have swept much anthropogenic air pollution away. Figure 2 shows the optical depths measured with a Volz sun photometer with wavelengths 328 lam and 500 gm. At the peak of the duststorm the optical depth was 1.8. There was no significant difference between the two wavelengths used, suggesting that the Angstrom exponent be near zero,
889
890
Parungo
Fig. 1. Concentrations of TSP and crustal elements (top), and of anthropogenic elements (bottom) during a duststorm.
Fig. 2. Optical depth measured during a duststorm.
Aeolian transport of Gobi dust and its radiation effects
891
and that the large particles attributed most scattering of solar radiation. Satellite data showed that the albedo over the dust plume doubled in channel 1 (0.58-0.68 pro) and increased 50% in channel 2 (0.731.10 gm), compared to the underlying background. These data suggest that increased dust in the air would increase backscatter of incoming solar radiation and reduce insolation on the surface. Indeed, the surface temperature records (Figure 3) show that on 12 April (the dustier day), the daytime maximum was 6~ to 10~ colder than the following days, while the nighttime minimum was only I~ to 3~ cooler. The fact that cooling during the day was greater than at night suggests that besides a cold-front invasion, dust attenuation of solar energy was acute. Apr. 12, 1994
Apr. 13, 1994
Apr. 15, 1994
Apr. 14, 1994
I
30~ o o
25'
\
20
E ~ 10 5 0
p
r
,
1 5
,
,
,
,
9 13 17 21
Hour
1
5
9
13 17 21
Hour
1
t
,
5
9
i
i
13 17 21
Hour
5
9 13 17 21 Hour
Fig. 3. Surface temperature during and after the duststorm. Dust particles that scatter solar radiation, can cool the earth's surface. They can also absorb solar radiation and warm the aerosol layers. Models have predicted that the temperature lapse rate should decrease in a duststorm. Figure 4 (fight) shows in the late afternoon only a general cooling, but no discernible difference in the lapse rate in the lower troposphere during a duststorm. The result suggested that the scattering effect overwhelmed the absorption effect. The dust's single scattering albedo was calculated to be 0.98, which also suggests a cooling effect. Figure 4 (left) shows the temperature profiles in the early morning after a dusty night (4/12) and a nondusty night (4/15). During nighttime when solar-radiation effect is absent, out-going terrestrial radiation dominates heat, budget. The dust particles which are generally large, can absorb the longwave terrestrial radiation and prevent heat to escape from below. As a result, nocturnal surface inversion was not observed in the dust night. A change of lapse rate between 3 km and 5.5 kin, where it probably contained a thick layer of duct, suggests that the layer acted like a blanket to keep the heat within. CONCLUSION This case study of a duststorm demonstrates that dust cools the lower troposphere and surface in the daytime. Because the dust's single scattering albedo was as high as 9.98, the dust scattering effect overwhelmed its absorption effect for solar energy. However, at night the dust layer can act as a thermal blanket, which reduces longwave terrestrial radiation escape from the earth's surface, and thus causes warming. The net radiative forcing of duststorms depends on the temperature changes for both day and night.
892
Parungo
Fig. 4. Profiles of dusty (4/12) and nondusty (4/15) conditions. REFERENCE Parungo, F., Kim, Y., Zhu, C., Harris, J., Schnell, R., Li, X., Yang, D., Feng, X., Zhou, M., Chert, Z. and Park, K. (1995 and 1996) Asian duststorms and their effects on radiation and climate. STC Tech. Report 2906, Part I (pp. 56) and Part II (pp. 80). Science and Technology Corporation, Hampton, VA.
AEROSOL OPTICAL DEPTH AND DIRECT-TO-DIFFUSE SOLAR IRRADIANCE RELATIONSHIPS IN THE NORTHEASTERN UNITED STATES Nels Laulainen* and Nels Larson* Qilong Min**, Joseph J. Michalsky**, Jim Schlemmer**, and Lee Harrison** *Pacific Northwest National Laboratory, Richland, Washington, USA-99352 **State University of New York at Albany, Albany, New York, USA-12205. Abstract - We have determined aerosol optical depths and direct-to-diffuse solar irradiance ratios over a four-year period from late 1991 through late 1995 at nine sites in the northeastern United States from observations made with a multi-filter rotating shadowband radiometer (MFRSR). The MFRSR measures total and diffuse irradiance on a horizontal surface in six narrowband wavelength intervals between 400 and 1000 nm. From these measurements we calculate the direct normal irradiance values, from which we derive the aerosol optical depth using the Beer-Lambert-Bougeur law and applying corrections for ozone absorption and Rayleigh scattering. We also calculate the direct-to-diffuse ratios. A radiative transfer model based on the adjoint method provides a means to calculate aerosol optical depth and surface albedo using the observed irradiance ratios. Using this model, we have examined the sensitivity of the results with respect to aerosol absorption index, aerosol phase function or asymmetry parameter and size distribution. From the observations alone, we notice the aerosol patterns at each site are consistent with an annual cycle superimposed on a decaying stratospheric aerosol loading from the Mt. Pinatubo eruption. The wavelength dependence of the aerosol optical depth shows changes in the aerosol size distribution. Keywords - aerosols, optical depth, solar irradiance, radiometry, radiative transfer INTRODUCTION In recent reviews, Charlson et al. (1992) and Penner et al. (1994) suggest that anthropogenic aerosols may exert a direct radiative influence on climate that is comparable in magnitude, but opposite in sign, to that of greenhouse gases. They also note that the magnitude of the effect may be uncertain by a factor of about 2. Much of the uncertainty in estimating the direct climate forcing effect of aerosols arises from the uneven distribution of aerosols over the planet (as a result of nonuniform aerosol precursor gas emissions, gas-to-particle conversion rates and aerosol scavenging and removal processes), as well as a wide range of values for aerosol mass light scattering efficiency. One of the parameters used in estimating the direct aerosol radiative forcing is the atmospheric aerosol optical depth (AOD), the integral over the vertical column of the aerosol light extinction coefficient. Aerosol light extinction as a function of height can be determined from direct observation with in situ instruments (such as an integrating nephelometer, an aerosol optical absorption device, a transmissometer, etc.) or derived from measured aerosol mass size distributions and chemical composition with the use of Mie theory. Altematively, AOD can be derived from direct solar irradiance measurements or from the application of a radiative transfer model using measured diffuse-to-direct irradiance ratios. The latter approach is the focus of this paper. The current study uses data from a network of 9 stations equipped with radiometric instrumentation located in the northeastern United States, known as the Quantitative Links Network (QLN), that operated from late-1991 until September 1995. This study was part of the Department of Energy's
893
Laulainen et al.
894
Quantitative Links Program, whose broad aims were to quantify linkages between changes in atmospheric composition and the temperature of the planet. The QLN was designed to collect data on the three components of solar irradiance -- total, diffuse, and direct -- at the earth' s surface. These data are being used to investigate how clouds and aerosol particles in the atmosphere affect the local energy budget and climate. In this paper, we present some highlights of our findings with respect to AOD and diffuse-to-direct irradiance ratios from a few selected sites of the QLN. We also compare these results with those obtained from a radiative transfer model. MEASUREMENTS AND INSTRUMENTATION The field measurements obtained from the QLN address the problem of relating changes in cloudiness and aerosol loading to observed solar irradiance at the surface and, ultimately, the changes in the amount of solar irradiance that is lost to space as a result of changing aerosol and cloud patterns. The principal instrument used to collect solar irradiance data is the Multi-Filter/Detector Rotating Shadowband Radiometer (MFRSR). A detailed description may be found in Harrison et al. (1994). Its design allows it to measure the total downwelling and diffuse solar irradiance simultaneously in 6 narrow spectral intervals (nominally with central wavelengths at 415, 500, 610, 665, 862 and 940 nm and bandwidth of 10 nm) and one broadband interval representing the total shortwave spectrum. From the two measurements of total and diffuse irradiance, the direct normal incident solar irradiance can be calculated. The direct beam data are corrected for the cosine response of the receiver disk. The direct beam data for cloud-free conditions are used through application of the Langley analysis method (regressing the natural logarithm of the direct irradiance against airmass) for determining the extraterrestrial solar irradiance Io and total optical depth at each wavelength. Frequent determinations of Io allow tracking of changes in Io caused by soiling and filter degradation. Until the advent of MFRSR-type techniques, determining the direct normal irradiance from the total and diffuse irradiation components was considered an unconventional approach to solar radiometry. Two important criteria for deriving direct normal irradiance values from such observations are that the receiving optics should have a cosine response and that deviations from true cosine response behavior be quantified. The MFRSR is certainly no exception. To test how well these criteria are met, simultaneous measurements with a standard MFRSR and a direct-beam module, consisting of an MFRSR detector head fitted with a baffled tube that blocks all but the direct irradiance and mounted on an automated tracker to follow the sun throughout the day, were performed. The results from the two types of measurements show that the rms errors in the total optical depth are of the order of 0.005 or less, indicating that the MFRSR-technique is fully comparable with traditional sun-tracking radiometers. Further details are found elsewhere (Harrison et al. 1994). DATA ANALYSIS Langley analysis is a method for simultaneously deriving mean optical depth 5 and the outside-theatmosphere solar intensity I o from observations of the direct beam intensity I during the course of a day in a given spectral band. The method is based on the Beer-Lambert-Bouguer law I = fSE ~ Io
~
e~'m,
where fSE is a factor to account for mean sun-earth distance and m is the relative airmass. The method involves regression of In I against m, and requires 5 to remain constant over the time span for which the observations are made. Optical depth retrieval via this method is complicated, however, by cloud transits and other time-varying interferences. We have developed a procedure (Harrison and Michalsky 1994) which objectively selects data points from a continuous time series and performs the needed regression. It is applicable to direct solar observations derived from MFRSR or other instruments that
Aerosol optical depth and direct-to-diffuse solar irradiance relationships
895
track the sun. The algorithm is capable of processing many data files and selecting the days for which the Langley regression can be performed. A comparison of optical depth calculated using a subjective algorithm and the automated technique when applied to a 1-year period of observations from Boulder, Colorado shows that the two methods are equivalent (Harrison and Michalsky 1994). The aerosol contribution to the total optical depth can be estimated by subtracting the contributions of molecular or Rayleigh scattering and any molecular absorption, from such species as ozone and water vapor, that may be present in a given wavelength band. Typically climatological values are used to make the Rayleigh and ozone corrections. Once the approximate wavelength dependence of AOD has been determined, it is possible to estimate more accurate AODs and ozone column abundances by using an iterative fitting procedure. The AOD has contributions from the troposphere and the stratosphere. The tropospheric component frequently exhibits a seasonal pattern, albeit with considerable variability. The stratospheric component is generally small compared to mean tropospheric values and has a much weaker seasonal pattern, with much less variability. However, occasional volcanic events can significantly perturb stratospheric aerosol loadings. Despite the large variability of the tropospheric component, it is possible to develop a smooth mean seasonal pattern (Michalsky et al. 1990) during periods of volcanic quiescence. When the seasonal tropospheric component is subtracted from the AOD time series, the residual time series show the occurrence of volcanic events. The Mt. Pinatubo eruption in 1990 and E1 Chichon eruption in 1983 caused major perturbations to the stratosphere. Occasionally, other events, such as wide-spread forest fires, can also be seen in such time series representations. For the purposes of this study, the daily average optical depth is calculated in two ways: 1) from the slope of the Langley regression line and, as a result of the many frequent determinations of Io, 2) from 30-minute averages calculated from the Beer-Lambert-Bougeur law using an estimated or interpolated median Io for that day (Michalsky et al. 1994). AODs for five wavelengths (the sixth narrowband wavelength at 940 nm is used to estimate water vapor) are obtained by subtracting Rayleigh scattering and Chappuis ozone absorption optical depths (climatological-average values) from the total optical depths, as indicated above. The AOD data are then displayed as a time series for each wavelength. Because of the large day-to-day variability, the time series is fit by a smoothing function procedure, called lowess (locally-weighted, robust regression). The ratio of diffuse-to-direct solar irradiance is a useful parameter for not only estimating certain aerosol optical properties, such as optical depth and light absorption index, but, when combined with wavelength-dependent attenuation, is a sensitive means of detecting and monitoring cirrus cloud frequency. King (1979), for example, demonstrated that the ratio of downwelling diffuse to direct irradiance in the mid-visible region can be related to independently observed values of AOD, ground albedo, and the aerosol light absorption index through Mie and radiative transfer theory. Optimum values of the surface albedo and the optical depth may also inferred from the diffuse-to-direct irradiance ratios using an inversion method that we have developed. The inversion method combines the use of an accurate and efficient adjoint/discrete ordinate method (Min and Harrison 1996) of radiative transfer and a nonlinear least squares procedure. The radiative transfer model includes all orders of multiple scattering and is valid for vertical inhomogeneous, nonisothermal, plane-parallel media. To account for vertical inhomogeneity of the various radiatively active gases and cloud/aerosol particles, and, therefore, optical properties such as single scattering albedo and phase function, the atmosphere is divided into a series of homogeneous layers in which the scattering and absorbing properties are constant within each layer, but may vary from layer to layer. The model is used to evaluate the diffuse-to-direct irradiance ratio and its derivatives. Because the diffuse-to-direct ratio is complicated by its dependence on the vertical distribution, size distribution, and refractive index of the
896
Laulainen et al.
aerosol, we adopt, as a first step, the MODTRAN2 aerosol model. The fitting procedure has been applied to data from several of the QLN sites. DISCUSSION The QLN has provided a unique set of radiometric and meteorological data to investigate how clouds and aerosols affect the local energy budget and climate. These data can be used to develop climatology of various directly observed variables, such as diffuse, direct and total irradiance, or computed parameters, such as the diffuse-to-direct ratio, surface albedo, and AOD as a function of wavelength. The data are also useful for examining time-series behavior of observed and computed quantities, such as AOD and cirrus cloudiness. We found, for example, that the AOD time series over the period 19921995 at each of the QLN sites is consistent with an annual cycle superimposed on a decaying aerosol loading associated with the Mt. Pinatubo eruption. By 1995 the effects of Mt. Pinatubo were minimal. Moreover, each site has a distinctive seasonal pattern. The seasonal changes are also reflected in the wavelength dependence of the aerosol time series, indicating seasonal changes in the aerosol size distribution. Results of the non-linear least squares procuedure coupled to the radiative transfer model produced consistent results for the AOD compared to those estimated from the direct beam. The time series for the inferred surface albedo appear to be consistent with other observed values. More work is under way to use the radiative transfer model in conjunction with the measured irradiance data and the AOD derived from the direct beam measurements to perform sensitivity tests on aerosol vertical distribution, aerosol size distribution and refractive index or other parameters, such as the single scattering albedo and phase function. Acknowledgement: This work is supported by the U.S. Department of Energy under Contract DEAC06-76RLO 1830. Pacific Northwest National Laboratory is operated for the U.S. Department of Energy by Battelle Memorial Institute. References Charlson, R.J., S.E. Schwartz, J.M. Hales, R.D. Cess, J.A. Coakley, Jr., J.E. Hansen, and D.J. Hofmann (1992) Climate forcing by anthropogenic aerosols. Science, 255, ~423-430. Harrison, L., and J. Michalsky (1994) Objective Algorithms for the Retrieval of Optical Depths from Ground-Based Measurements. Applied Optics, 33, 5126-5132. Harrison, L., J. Michalsky and J. Berndt (1994) Automated Multifilter Rotating Shadowband Radiometer: An Instrument for Optical Depth and Radiation Measurements. Applied Optics, 33, 51185125. King, M.D. (1979) Determination of the Ground Albedo and the Index of Absorption of Atmospheric Particulates by Remote Sensing. Part II: Application. J. Atmos. Sci., 36, 1072-1083. Michalsky, J.J., E.W. Pearson, and B.A. LeBaron (1990) An Assessment of the Impact of Volcanic Eruptions on the Northern Hemisphere's Aerosol Burden During the Last Decade. J. Geophys. Res., 95D, 5677-5688. Michalsky, J.J., J.A.Schlemmer, N.R. Larson, L.C. Harrison, W.E. Berkheiser III, and N.S. Laulainen (1994) Measurement of the Seasonal and Annual Variability of Total Column Aerosol in a Northeastern U.S. Network. In Proceedings of the International Specialty Conference on Aerosols and Atmospheric Optics." Radiative Balance and Visual Air Quality, Snowbird, Utah, September 26-30, 1994, pp. 247258, Air and Waste Management Association, Pittsburgh, Pennsylvania. Min, Q.-L., and L.C. Harrison (1996) An Adjoint Formulation of the Discrete Ordinate Radiative Transfer Method. J. Geophys. Res., 101D, 1635-1640. Penner, J.E., R.J. Charlson, J.M. Hales, N.S. Laulainen, R. Leifer, T. Novakov, J. Ogren, L.F. Radke, S.E. Schwartz and L. Travis (1994) Quantifying and minimizing uncer-tainty of climate forcing by anthropogenic aerosols. Bull Amer. Meteorol. Soc., 75, 375-400.
MEASUREMENTS OF HYGROSCOPIC GROWTH OF ATMOSPHERIC SUBMICROMETER PARTICLES DURING A TRANSECT OF THE PACIFIC OCEAN RADOVAN KREJCI*, OLLE BERG, ERIK SWIETLICKI. Division of Nuclear Physics, Lund University, S/31vegatan 14, 223 62 Lund, Sweden Czech Geological Survey, K15rov 3/131, 118 21 Praha 1, Czech Republic .1:
Abstract - A preliminary evaluation of the measurements of hygroscopic growth of submicrometer aerosol particles during the first leg of the ACE-1 experiment (Oct. 11 - Nov. 10 1995) is presented. The hygroscopic growth was measured with a H-TDMA (Hygroscopic Tandem Differential Mobility Analyser) on board the NOAA research ship R/V Discoverer during the north-south Pacific Ocean cruise between Seattle, USA and Hobart, Tasmania, Australia. The data presented here were taken between positions 33 ~ N . . . . . 5 ~ W and 40 ~ S, 156 ~ E. Measurements were performed in aged remote marine tropospheric aerosols as well as during periods with air masses showing a clear anthropogenic inlluence.
K e y w o r d s - Hygroscopic growth" Tandem differential mobility analyser; Submicrometer atmospheric aerosol; Pacific Ocean.
INTRODUCTION The Intergovernmental Panel on Climate Change (IPCC) clearly recognises the importance of atmospheric aerosols in the global radiative balance (IPCC, 1994). While the combined positive climate forcing of the greenhouses gases can be fairly well estimated, the negative forcing due to atmospheric aerosols is highly uncertain and currently poses the largest source of uncertainty in predictions of the future global climate. Atmospheric aerosols can affect the radiative balance in two ways. Submicron particles back-scatter incoming short-wave solar radiation efficiently and thereby increase the global albedo. This is the so called direct effect. The aerosol particles containing the largest amounts of hygroscopic material on an individual particle basis can act as cloud condensation nuclei (CCN) and thus produce an indirect negative radiative forcing, since an increased number of CCN tends to increase the albedo of the marine stratocumulus clouds. The average cloud life time is probably also prolonged since smaller cloud droplets are less prone to form precipitation. Evidently, the hygroscopic properties of the aerosol are important for both the direct and indirect forcing. A series of international aerosol characterisation experiments has been planned in order to elucidate the chemical and physical processes controlling the evolution and properties of atmospheric aerosols and their role in radiative climate forcing. The first in this series of experiments, the Southern Hemisphere Marine Aerosol Characterization Experiment (ACE-1) was staged in the area south-west of Tasmania, Australia between Nov. 15 and Dec. 14 1995 (Bates et al., 1995). The Tasmanian Sea was chosen for the ACE-1 experiment as it offered a possibility to study the atmospheric aerosol in an unpolluted, remote marine environment. The measurements of the hygroscopic growth of submicrometer aerosol particles were perforIned on the NOAA research vessel R/V Discoverer. This presentation covers the observations made during the
897
898
Krejci et al.
north-south transect across the Pacific Ocean flom the NOAA home port in Seattle, USA to the ACE-1 base port in Hobart, Tasmania, Autralia. EXPERIMENTAL The H-TDMA measurements presented here were conducted on board a ship, the NOAA R/V Discoverer, during a north-south transect across the Pacific Ocean from Seattle, USA to Hobart, Tasmania, Australia between Oct. 15 and Nov. 10 1995. The transect was a preparation for the ACE-1 intensive campaign. The data were taken between positions 33 ~ N, 144.5 ~ W and 40 ~ S, 156 ~ E. The ambient air was sampled through a whole-air inlet and transported into the laboratory container housing the aerosol equipment. The intrument used to measure the hygroscopic growth was the HTDMA which consists of two Differential Mobility Analysers (DMA) in series separated by a relative humidity (RH) conditioning unit. DMA-I selects a narrow, nearly monodisperse size range of the atmospheric aerosol at dry conditions (RH < 10%). The aerosol is then humidified to a RH of 90% and sized with DMA-2. The hygroscopic growth factor is defined as the particle diameter at 90 % RH divided by the particle diameter at a RH < 10%. RESULTS During the Pacific Ocean transect, the hygroscopic growth of aerosol particles was investigated at four dry particle sizes. These were; two in the Aitken mode (35 nm and 50 nm diameter), one in the accumulation mode (165 nm) and one at an intermediate size (75 nm). In continental polluted aerosols, a bimodal hygroscopic behaviour is normally seen (e.g. Svenningsson et aI., 1994, 1996). The two groups of particles with different hygroscopic growth are denoted lesshygroscopic and more-hygroscopic respectively. In such anthropogenically influenced aerosols, the growth factors even for the more-hygroscopic group were always less than those expected for pure salts of the major ions making up the aerosols. This discrepancy can be accounted for by attributing a rather large water-insoluble volume fl'action of around 50% to each of the particles in this group. During the Pacific Ocean transect, only one group of particles showing similar hygroscopic behaviour was observed in a majority of cases for all four dry sizes. The growth factors were generally higher, or even much higher than for anthropogenically influenced aerosols (Fig. I). This more-hygroscopic group had growth factors between 1.36-1.66 for 35 nm dry size, 1.42-1.68 for 50 nm dry size, 1.49-1.69 fox 75 nm dry size and finally 1.52-1.75 for 165 nm dry size. Assuming that the soluble part of the aerosol consisted entirely of ammonium sulphate, then these growth factors would correspond to soluble volume fractions (for individual particles) ranging from 0.35-0.94. The stability of the hygroscopic behaviour is an indication that the marine aerosol in general had a long residence time within a wellmixed marine boundary layer. The occurrence of a group of less-hygroscopic particles could be attributed to anthropogenic influence. Most of these comparatively rare occasions were observed either on the northern hemisphere between 40 ~ and 20 ~ N, or in the southern hemisphere between 35 ~ and 40 ~ S, and were the result of outbreaks of pollution flom the North American continent, the eastern coast of Australia or New Zealand. Apart from these two main anthropogenically influenced regions of the Pacific Ocean, a lesshygroscopic group of particles could be observed in the vicinity of the Hawaii islands and Rarotonga in the Cook Islands. The less-hygroscopic group was never observed for the accumulation mode particles (at 165 nm) but only fox" the Aitken mode particles (35 and 50 nm and in some cases also for 75 nm dry size). The average hygroscopic growth factor fox this less-hygroscopic group varied, when observed, between 1.15-1.42 for 35 nm dry size, 1.15-1.43 fox 50 nm dry size and 1.20-1.45 for 75 nm dry size particles.
Measurement of hygroscopic growth of atmospheric submicrometerparticles
899
Figure 1 gives a compilation of the H-TDMA data for aerosol particles in the more-hygroscopic group obtained during the Pacific Ocean transect. The growth factors are given for twelve various periods (latitudes) along the cruise during which measurements could be performed under fairly stable atmospheric conditions. The points represent the arithmetic average of the growth factors measured around the position in question with a measure of the variation in the growth factors given as one standard deviation.
35 nm d r y p a r t i c l e 1.75
- ,
,
-
_,
50 nm d r y p a r t i c l e
size ,
-
,
,
1.75
,
1.7
1.7 0 o
1.65
,4-
1.6
. E
0
:{:
G)
E 13
.
.
,-
-,
.
.
.
E
1.45 1.4
13
1.45
-
1.4
_
.~_
'
'-
'
i
'
0
165 nm d r y
size
9
50
)article
size
1.75
r
1.7
117
1.65
1.65 .t:z
1.6
1.6
1.55
1.55 1.5
. . . . .
1.45 1.4
.
Latitude
0
a) E
-,
.
- -
-50
50
75 nm d r y p a r t i c l e
L
.
i
1.35 0
1.75
0
.
1.5 Q
Latitude
o
.
0
1.55
-50
..~ ,,i,-,
.
t,-
1.35
~_
size
9- . . . . . .
0 o
::
1.5
. E
..,
. .
.
. .
.
.
.
.
.
"-
1.5
I::
1 .45
.~_ 13
.
,
-50
0
Latitude
50
,
1.4 1.35
1.35
i
,
50
I
0
50
Latitude
Figure 1. Hygroscopic growth factors measured by the H-TDMA during the north-south Pacific Ocean transect between Oct. 15 and Nov. 10 1995. The growth factors represent the relative change ill particle diameter when taken from a dry state to a relative humidity of 90 %. The bars indicate one standard deviation for the measurements performed at the location in question.
For all four dry sizes, the lowest growth factors can typically be observed at the beginning and the end of the cruise, where anthropogenic pollutions from North America and Australia has a significant influ-
900
Krejci et al.
ence. A different pattern can be seen in the tropics around the Inter-Tropical Convergence Zone (ITCZ) located between 5 ~ N and 10" S. Here, the growth factors of particles in the more-hygroscopic group were very stable with time and latitude. The stability of the aerosol and the fairly high average growth factors of 1.57 for 35 nm, 1.61 for 50 nm, 1.62 for 75 nm and 1.66 for 165 nm are indications of a long tropospheric residence time and almost no influence from anthropogenic sources. The hygroscopic properties south and north of the ITCZ show a more varied behaviour and reflect different general atmospheric air mass movements such as the presence of cyclonic and anti-cyclonic systems and frontal passages. Moreover, the influence tiom anthropogenic sources was sometimes evident. The average growth factors in these regions varied between 1.52-1.62 for 35 nm particles, 1.541.65 for 50 nm particles, 1.54-1.67 for 75 nm particles and 1.57-1.72 for 165 nm particles. The hygroscopic growth factors were not substantially lower in the more polluted Northern hemisphere compared to the Southern hemisphere. A more detailed analysis involving also other observed quantities such as aerosol size distributions and aerosol chemical composition coupled with air mass back trajectories will be undertaken in order to fully explain the hygroscopic properties of the submicrometer aerosol particles. CONCLUSIONS The presented data is the result of a preliminary evaluation of the H-TDMA data obtained on the Pacific Ocean north-south transect during the first leg of the ACE-1 experiment. The measurements were performed in both anthropogenically influenced air masses and remote marine tropospheric aerosols around the ITCZ. The occurrence of a less-hygroscopic group of particles could be attributed to anthropogenic sources. The growth factors were less than what is expected from aerosol particles consisting entirely of a solution of the major ions constituting the aerosol, although the water insoluble volume fiaction (on an individual particle basis) was considerably smaller for the Pacific Ocean aerosol than what was previously observed in polluted continental aerosols. The hygroscopic growth factors were not found to be substantially lower in the more polluted Northern hemisphere compared to the Southern hemisphere. The observations on the hygroscopic behaviour of the remote marine tropospheric aerosol serve as vital input data for models describing the radiative effect of remote marine aerosols.
REFERENCES Bates, T. Gras, J. and Huebert, B. (1994)ACE-1 Science cmd bnpIementation Plan. IPCC (Intergovernmental Panel on Climate Change), Climate Change 1994, Radiative Forcing of Climate Ctumge, Cambridge University Press, 1994. Berg O., Krejci R. and Swietlicki E. (1996) Changes in hygroscopic growth of atmospheric submicrometer particles during air mass subsidence events in remote marine environments, These proceed-
ings. Svenningsson B., Hansson H.-C., Wiedensohler A., Noone K.J., Ogren J.A., Hallberg A. and Colvile R. (1994) Hygroscopic growth of aerosol particles and its influence on nucleation scavenging in cloud: Experimental results from Kleiner Feldberg, J. Atmos. Chem. 19, 129-152. Svenningsson B., Hansson H.-C., Martinsson B., Wiedensohler A., Swietlicki E., Cederfelt S.-I., Wendisch M., Bower K.N., Choularton T.W. and Colvile R.N. (1996) Cloud droplet nucleation scavenging in relation to the size and hygroscopic behaviour of aerosol particles, Atm. Env. In press.
INVESTIGATION MATURE
OF AEROSOL
CONVECTIVE
-
DROPLET
INTERACTION
IN THE
CLOUDS USING THE TWO-DIMENSIONAL MODEL.
E.N.STANKOVA, M.A.ZATEVAKHIN International Institute for Interphase Interactions, St. Petersburg, Russia. Abstract - Two-dimensional cloud model was used for the investigation of aerosol - cloud droplet interaction in mature convective cloud, forming as a result of catastrophic natural and anthropogenic phenomena. The results of numerical simulation showed high efficiency of the process of cloud droplet - giant aerosol particles interaction (so called coalescence nucleation effect) on scavenging of large scale aerosols. K e y w o r d s - Aerosol -droplet interaction; Convective cloud; Coalescence nucleation effect INTRODUCTION Interaction of cloud drops with aerosol particles in the mature convective clouds, forming as a result of catastrophic natural and anthropogenic phenomena such as accident industry pollutants, explosions, large forest fires, volcanoes eruptions is of great scientific interest in connection with the role it plays in transport, transformation and scavenging of aerosols, forming under such extreme conditions. This problem was investigated by means of numerical simulation in the works of Bradley (1987), Cotton (1984), Giorgi (1989), Penner and Haselman (1986), Molenkamp (1980), Carhart and Policastro (1988). Numerical models used in these works have aerosol blocks developed with different state of detailing. The most elaborated schemes of aerosol - droplet interaction are presented in the works concerning smoke particles scavenging within plumes, forming above large fires (Bradley (1987) and Cotton (1984)), the most simple ones - in the works, concerning wet fallout from bubbles of hot air produced by explosions (Molenkamp (1980), Carhart and Policastro (1988)). In this study we investigate cloud - droplet interaction in the mature cloud, forming as a result of hot air rising over the momentary high energy source, but unlike Carhart and Policastro (1988) we use rather elaborated aerosol block, similar to that, presented in the work of Bradley (1987). Besides, dynamic block of our model is appropriate for the description of the flows with arbitrary vertical scales and temperature variations and so reflects the specific features of the clouds developing in extreme conditions more adequately. DESCRIPTION OF THE MODEL The model is time-dependent, axisymmetric, based upon the solution of equations, describing the turbulent flow of compressible two-phase two-component medium. Equations of the model were obtained with the following assumptions: all components have the same temperature and move with the same velocity with the exception of rain drops which vertical speed deviates from the velocity of the flow at a value equal to their terminal velocity. Microphysical block of the model addresses only warm rain processes. The parameterization of the moist processes are identical to those used in the Shiino (1977) cloud model. Detailed description of the model is presented in the works of Stankova (1994) and Dovgalyuk et al. (1994). Three fractions of aerosol particles are considered in aerosol block of the model: dry aerosol, aerosol in cloud droplets and aerosol in raindrops. Dry aerosol particles are divided in two parts: small and large aerosol particles. Small particles are entirely involved in the motion of outer flow and large 901
902
Stankova and Zatevakhin
particles are falling with terminal velocities dependent on the their size. It is assumed that the influence of aerosol particles on dynamic processes is negligible in contrast to microphysical processes. Mutual transitions of aerosol from one category to another take place by means of the processes of condensation nucleation (small dry particles transform to aerosols in cloud droplets), of so called coalescence nucleation (large aerosol particles transfer to the raindrops while colliding with cloud drops during its fall), of autoconversion and accretion (aerosol in cloud drops transform to the aerosol particles in rain drops in the process of rain formation) and of evaporation (aerosol in cloud and raindrops resuspend in the air if the drop evaporates). Condensation nucleation rate is calculated by the formulas used by Bradley (1987). The equation for the rate of coalescence nucleation is derived under the assumption that terminal velocity of large aerosol particles is proportional to its linear size and that the collection efficiency is equal to unity. SIMULATION RESULTS Formulation of the problem and temperature stratification are the same presented in the work of Dovgalyuk etal. (1994) with the exception of the initial values of the height of thermal's centre and temperature, which are taken here equal to value of the radius and 2000 K consequently. The relative humidity profile is defined as suggested by Carhart and Policastro (1988) for the case of 90% value. The results of numerical simulations show that when rising the thermal transforms into a toroidal vortex, the temperature in which gradually decreases while rising. The air cooling results in the beginning of the condensation process near the axis of the symmetry, then the process propagates to the whole cloud volume, including vortex core. While further rising thermal reaches upper, more dry layers of the atmosphere and its liquid water content decreases. By this time however condensation process begins in the updraught, similar to the vertical flows in natural convective clouds. These processes are discussed in detail by Stankova (1994). The analyses of the results shows that quality structure of the distribution of aerosol particles of all classes in space and time depend upon the total aerosol mass and concrete parameters of aerosol size distribution only slightly. Figure 1 shows the typical picture of aerosol mass concentrations distribution in space at time moment equals to 6 rain, when the thermal has already approached maximum height of its rising but the intensive circulation flow is still preserved in the vortex core.
Fig 1. Space distributions of: small dry aerosol particles (a), aerosol particles in cloud drops (b), large dry aerosol particles (c), aerosol particles in raindrops (d).
Investigation of aerosol-droplet interaction in the mature convective clouds
903
As it can be seen from fig. la the distribution of small dry aerosol particles has rather complex structure. High concentrations of small dry aerosol particles locate in the regions of cloud droplet evaporation, where circulation flow round vortex core takes place. Regions of low dry aerosol particle concentration locates near cloud lower boundary and under the vortex core, where the ambient air is involved into the cloud and where the nucleation process takes place. On the contrary, maximum concentrations of aerosols in cloud droplets one can notice inside the vortex and near lower boundary of the cloud (see fig. lb). Distributions of the mass concentrations of large dry aerosols and aerosols in raindrops presented in fig. lc and l d consequently, demonstrate the role of the coalescence nucleation process in aerosol transformations. Large aerosol particles can sweep out cloud droplets during their fall through a cloud so effectively, that large dry aerosols are presented only in the regions of dry air and aerosols in raindrops - everywhere in the presence of any amount of condensed water. But it must be noted that the higher is aerosol particle concentration the thinner is water film particle is covered by. Numerical simulation results show also that about 90% of large dry particles and about 50% of small dry particles transfer to the cloud and raindrops. These values are essentially greater then that obtained by Carhart and Policastro, where coalescence nucleation process was not considered. It must be noted that the last value can be essentially varied by more accurate definition of condensation nucleation process calculation, taking into account concrete value of nucleation activity of small aerosol particles, depended upon their physical and chemical features. The first value seemed to be much better founded, because the rate of coalescence nucleation process is 1.5 s-I for the particles of 100 tz diameter and it increases proportional to the cube of particle size. That is why the rate of coalescence nucleation can increase by the several orders of magnitude in the case when the spectrum of large aerosol particles is sufficiently wide. This rate is so large that even more accurate definition of the collection efficiency coefficient cannot change these results qualitatively. REFERENCES Bradley, M.M. (1987) Numerical simulation of nucleation scavenging within smoke plume above large fires. Proceedings Int. Conf. Energy Transform., Lausanne, Switzerland, March 2-6. Carhart, R.A. and Policastro, A.J. (1988) Effects of relative humidity and yield on self-induced rainout from tactical nuclear explosions. Simulation, November, 191-194. Cotton, W.R. (1984) A simulation of cumulonimbus responses to large fires storm. Implication to the nuclear winter. Proceedings of the 9-th Int. Conf. of Cloud Physics. 4, 927-932. Dovgalyuk, Y.A. Zatevakhin, M.A. and Stankova, E.N. (1994) Numerical simulation of a buoyant thermal using the k-e model. J.Appl.Met. 33, 1118-1126. Giorgi, F. (1989) Two-dimensional simulations of possible mesoscale effects of nuclear fires. J. Geophys.Res. 19, D1. Molenkamp, C.R. (1980) Numerical simulation of self-induced rainout using a dynamic convective cloud model. Proceedings VIII International Conference on Cloud Physics. Clermont-Ferrand, France, July 15-19, 503-506. Penner, J.E. and Haselman, L.C. (1986) Smoke plume distribution above large scale fires. Implications for simulations of "Nuclear Winter". J.Clim.Appl.Met. 28, 1434-1444. Shiino, J. (1978) A numerical study of precipitation development in cumulus clouds. Pap.Meteor.Geophys. 29, 157-193. Stankova, E.N. (1994) Ph.D. Dissertation. Main Geophysical Observatory, St. Petersburg, Russia.
THE GLOBAL DISTRIBUTION OF SULFATE AEROSOLS CALCULATED WITH THE GRANTOUR/ECHAM COUPLED MODEL C. R. MOLENKAMP, J. E. PENNER, J. J. WALTON, C. J. O'CONNOR Atmospheric Sciences Division, Lawrence Livermore National Laboratory P.O. Box 808, Livermore, CA 94550, USA Abstract - We have recently coupled our tropospheric chemistry and transport model, GRANTOUR, with the ECHAM global climate model. With the coupled model we have estimated the global sulfur distributions over an annual cycle. Spatial distributions from these simulations will be compared to observations and to our earlier simulations. We will also calculate the climate forcing due to anthropogenic sulfate. Keywords - Aerosol; Climate forcing; Global climate; Precipitation scavenging INTRODUCTION Atmospheric aerosols are important to many processes within the atmosphere. They are important to radiation transfer and global climate. Aerosols acting as condensation nuclei control the microphysical and optical properties of clouds. The life cycle of many atmospheric trace gases and pollutants are highly dependent on interactions with atmospheric aerosols and subsequent removal by precipitation scavenging. Observation of the atmosphere by lidar and satellites often depends on scattering from atmospheric aerosols, and aerosols affect atmospheric visibility and the electromagnetic extinction of radar and lidar. The interactions between aerosols and other atmospheric constituents and properties are complex and diverse. While dust, sea-salt particles, inorganic carbonaceous aerosols, and biomass material may be important in some situations and locations, sulfate aerosols usually have the largest impact. Therefore, knowledge of the spatial distribution of natural and anthropogenic sulfate aerosols is important. The effects of sulfate aerosols on global climate can be both direct by scattering of solar radiation and indirect through alteration of the cloud droplet size distribution which enhances the back-scattering of solar radiation by clouds and perhaps alters the life cycle of clouds. The resultant negative climate forcing is significant and tends to counteract the warming associated with increased greenhouse gases (Charlson, et al., 1992; Kiehl and Briegleb, 1993). However, since the distribution of sulfate aerosols is regionally inhomogeneous, the pattern of forcing is quite different from greenhouse gas warming. Furthermore, changes induced in the cloud droplet size distribution can affect global temperature and precipitation patterns producing feedback between aerosols and climate (Penner, et al., 1993c). GRANTOUR We have previously examined the climate effects of sulfate aerosols and greenhouse gas forcings using our tropospheric chemistry and transport model (GRANTOUR) in conjunction with CCM1, and concluded that representation of the regional distribution of atmospheric aerosols was essential for reliable prediction of climate change (Taylor and Penner, 1994). We have also used GRANTOUR to simulate the global transport and deposition of 222Rn and 21~ (Dignon, et al., 1993), 0 3 and OH (Atherton, 1993; Atherton et al., 1993; Penner et al., 1993a), organic nitrates (Atherton, 1989),
904
Global distribution of sulfate aerosols calculated with GRATOUR/ECHAM model
905
anthropogenic aerosols (Taylor and Penner, 1994), smoke (Ghan et al., 1988), soot aerosols from biomass burning (Penner et al., 1991b), and black carbon (Penner et al., 1993b). And we have used GRANTOUR to simulate the global nitrogen budget (Atherton et al., 1991; Penner et al., 199 l a) and the global sulfur cycle (Erickson et al., 1991). All these simulations used CCM1 to provide global flow fields and meteorological data. COUPLED MODEL We have recently coupled GRANTOUR with the ECHAM global climate model (Deutsches Klimarechenzentrum, 1993) which provides several enhanced capabilities in the representation of aerosol interactions. Besides a better representation of global climate, ECHAM includes vertical mass fluxes (up and down) associated with cumulus convection, precipitation production and evaporation associated with convection, and large-scale cloud fractions, cloud water mixing ratios, and precipitation rates. The specific representation of cloud liquid water facilitates an improved representation of wet phase gas-to-particle conversion of SO2 to SO 2-. The large-scale cloud fractions and precipitation rates permit an improved parameterization of precipitation scavenging by stratiform clouds. And the convective mass fluxes and precipitation rates allow for an improved representation of mixing and scavenging by convective clouds. The current ECHAM/GRANTOUR model is only coupled in one direction. ECHAM is used as a meteorological driver for GRANTOUR. This has the advantage of allowing us to run ECHAM once to generate the meteorological data and then developing parameterizations for GRANTOUR with a constant set of meteorology. It has the disadvantage of not allowing feedback of the aerosols on the evolution of the weather and climate system. We generated one year of meteorological data from ECHAM at T21 resolution. Four hour averages of the 12 variables in Table 1 were saved and constitute the meteorological data set; all are three-dimensional variables defined on the ECHAM grid. GRANTOUR interpolates the variables to a constant sigma vertical coordinate that corresponds with the ECHAM variable grid for a surface pressure of 10SPa. Both models use the same horizontal grid. Vertical velocity is derived from U and V using the continuity equation. The parameterizations of large-scale scavenging and of convective mixing and scavenging are actually performed on the ECHAM vertical grid with trace species mixing ratios interpolated from the GRANTOUR grid (Molenkamp, et al., 1995).
Table 1. ECHAM Variables Passed to GRANTOUR Name U V T q,. ql
Variable
Units
Zonal Velocity
km/hr
Meridional Velocity
km/hr
Air Temperature Water Vapor Mixing Ratio
K
Pt.s
Liquid Water Mixing Ratio Large-Scale Precipitation Rate
cm/hr
Pcv
Convective Precipitation Production Rate
cm/hr
M,, M.
Convective Mass Flux Up
kg/m2/s
Convective Mass Flux Down
kg/m2/s
ph
Half-Level Pressure
Pa
F~ ICz
Large-Scale Cloud Fraction Vertical Diffusion Coefficient
m2/s
906
Molenkamp et al.
In our simulations, anthropogenic sulfur is supplied as SO2 by fossil fuel combustion, industrial sources, and biomass burning; natural sources include biogenic oceanic sulfate (DMS), terrestrial soils (DMS and HzS ), and vegetation (H2S). Gas phase interactions between these constituents and spatiallyand seasonally-varying background concentrations of OH and 03 in conjunction with wet-phase production inside clouds leads to the formation of sulfate particles. Removal of the sulfate particles occurs through precipitation scavenging and dry deposition. In the future we will provide four hour average trace species mixing ratios and/or aerosol optical properties to ECHAM. We will then run the models fully coupled, permitting us to evaluate some of the feedback effects of sulfate aerosols. SIMULATIONS With the coupled model we are calculating the global sulfur distribution over an annual cycle. We include mixing ratios for background DMS/H/S, SO 2, and SO4z- and for anthropogenic SO 2 and SO4z-. All the sulfate is assumed to exist in soluble aerosols. The simulations are currently in process and we will be presenting spatial distributions from these simulations and comparing them to observations and to our earlier results. We also intend to calculate the climate forcing due to anthropogenic sulfate. ACKNOWLEDGEMENT This work was performed under the auspices of the U. S. Department of Energy by the Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48. REFERENCES Atherton, C. S. (1989) Geophys. Res. Lett., 16, 1289-1292. Atherton, C. S., J. E. Penner, and J. J. Walton (1991) LLNL Report UCRL-JC-107223. Atherton, C. S. (1993) Ph. D. Dissertation, University of California, Davis. Atherton, C. S., J. Dignon, J. E. Penner, S. SiUman, and J. J. Walton, (1993) Presented at the IAMAPIAHS International Scientific Meeting, Yokohama, Japan, July 11-23. Charlson, R. J., S. E. Schwartz, J. M. Hales, R. D. Cess, J. A. Coakley. Jr., J. E. Hansen, and D. J. Hofmann (1992) Science, 255, 423-430. Deutsches Klimarechenzentrum (1993) The ECHAM 3 Atmospheric General Circulation Model. Report No. 6, Hamburg. Dignon, J., J. E. Penner, and J. J. Walton (1993) LLNL Report. Erickson III, D. J., J. J. Walton, S. J. Ghan, and J. E. Penner (1991) Atrnos. Environ., 25A, 2513-2520. Ghan, S. J., M. C. MacCracken, and J. J. Walton (1988) J. Geophys. Res., 93, 8315-8337.. Kiehl, J. T., and B. P. Briegleb (1993) Science, 260, 311-314. Molenkamp, C. R., J. E. Penner, J. J. Walton, and C. J. O'Connor (1996) Proceedings International Cloud Physics Conf., Zurich, August 19-23. Penner, J. E., C. S. Atherton, J. Dignon, S. J. Ghan, J. J. Walton, and S. Hameed (199 l a) J. Geophys. Res., 96, 959-990. Penner, J. E., S. J. Ghan, and J. J. Walton (1991b) In Global Biomass Burning, Ed. J. Levine, MIT Press, Cambridge, MA, 387-393. Penner, J. E., C. S. Atherton, and T. Graedel (1993a) In Global Atmospheric-Biospheric Chemistry: The First IGAC Scientific Conference, OHOLO Conference Series Books, Plenum Publishing, New York. Penner, J. E., H. Eddleman, and T. Novakov (1993b) Atmos. Environ., 27A, 1277-1295. Penner, J. E., R. J Charlson, J. M. Hales, N. Laulainen, R. Leifer, T. Novakov, J. Ogren, L. F. Radke, S. E. Schwartz, L. Travis (1993c) Quantifying and Minimizing Uncertainty of Climate Forcing by Anthropogenic Aerosols, DOE/NBB-0092T. Department of Energy, Washington, D.C. Taylor, K. E. and Penner, J. E. (1994) Nature, 369, 734-737.
CHARACTERISTICS OF THE AEROSOL OPTICAL DEPTH UNDER THE CLASSIFICATION OF THE SURFACE VISIBILITY F A N G LI and D A R E N LU
LAGEO, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China Abstract - The observation of direct solar radiation in urban district of Beijing was carried out with a PIS spectrometer from March 1993 to March 1995. The aerosol spectral optical depths are deduced from the radiation data. The results show that although seasonal mean aerosol optical depth is larger in spring and summer, smaller in autumn and winter, the aerosol spectral optical depth is closely related to surface visibility. For the same visibility level, it has very weak seasonal variation. This means the seasonal variation of the aerosol spectral optical depth is mainly determined by the variation of surface visibility. The yearly averaged aerosol optical depth spectra under the various visibility levels are given and the comparison with the calculation from an aerosol model of LOWTRAN 7 are made at the same visibility level. Keywords - Visibility; Aerosol optical depth; Radiation transfer. INTRODUCTION The potential effects of increased human activities on climate and air quality have prompted increased efforts to characterize properties of the globe atmosphere during the past few decades (Charlson, et al., 1992). Much of the impetus for this work has been stimulated by the conspicuous presence and observed impact of anthropogenic emissions associated with many urban environments. The optical properties of aerosol particles, expressed in terms of optical depth, determine radiation transfer and visibility in the atmosphere. Methods are needed to relate the optical properties of atmospheric particles to the meteorological factors, for example the surface visibility. Although there are visibility assessment methods to manage air quality and visual distance in urban areas such as Houston, TX (Dzubay et al., 1982), Los Angels, CA ( Cass, 1979), and Denver, CO (Sloane et al., 1991). Those were of relatively few attentions on the relationship of aerosol optical depth to visibility. The objectives of this paper are: (1) to investigate the optical properties of aerosols observed under the classification of the surface visibility; and (2) to compare the observed results with the calculated with urban aerosol model of LOWTRAN 7 code (Kneizys, et al., 1988) for displaying the deviation extent. EXPERIMENTAL A solar spectrometer of the PIS-A type is used to obtain aerosol optical depths. It is a portable instrument with the 512 silicon array sensing elements. The wavelength range is from 0.40 to 1.04/zm. The spectral resolution is 1.25nm. The spectral-response time is 10ms. The reproducibility error is less than 2%. The Langley method is adopted for the calibration of the spectrometer. The calibration was carried out under the stable and fine weather, usually in October of a year. The air mass dependence of
907
Li and Lu
908
the direct solar radiation is nearly of a perfect line on a logarithm basis at the wavelength of 0.55/.tm. The fluctuating error is within the 3%. Continual ground observations were carried out from March 1993 to March 1995 at the top platform of main building in Institute of Atmospheric Physics, Beijing, China. The days that were clear and part cloudy would be chosen for the observation. All of measurements were carried out when there were no clouds in the visual field of 5 o around the solar disk. The observation time was decided usually in the noon period 11:00-13:00. There are 1064 pieces of the data available for the analyses. Determination of the surface Visibility follows the instruction of visibility inspecting recommended by WMO's Committee of Instruments and Methods of Observation (CIMO) and the gradation of visibility is shown in table 1. After total atmospheric optical depths are inferred from the solar direct radiation measured on the ground, aerosol optical depth can be achieved with the reduction of gaseous molecule scattering and absorption from the total optical depth. Table 1. Classification of the surface visibility Visibility l];rade Visual range (km)
II 5
I 2
III 10
IV 20
30
RESULTS AND DISCUSSION There were research reports about statistical results of aerosol optical depth that show strongly seasonal variation in the random visibility (Qiu, J. H., et al., 1986). When the observation results are classified in terms of the grade of the surface visibility in table 1, the season variation of aerosol optical depth largely declines and the fluctuate range is approximately similar to the magnitude of observation error. The results are shown in table 2 at the wavelength of 0.55pm in urban area of Beijing in 1994, where [A I is mean absolute error, 5 (%) mean relative error, and 5,ax(%) a maximum relative error. Table 2. Under the classification of visibility the aerosol optical depths at the wavelength of 0.55ktm in Beijing area in 1993 and 1994. Visibility ~rade Season Spring Summer Autumn Winter Averase
IAI 5 (%)
I
1.6578 1.5541 1.5526 1.5882 0.0464 2.9 4.4
II
III
IV
V
1.0310 0.9457 0.9826 0.9075 0.9667 0.0401
0.6574 0.7208 0.5850 0.6128 0.6252 0.0451
0.4790 0.4347 0.4614 0.4293 0.4511 0.0191
0.3016 0.3761 0.3383 0.3389 0.0253
4.1 6.7
7.0 12
4.2 6.2
7.5 11
Characteristics of aerosol optical depth under classification of surface visibility
909
The spectral optical depths of atmospheric aerosols behave in the same way. Fig. 1 gives the results averaged for two years of 1993 and 1994 under the gradation of the surface visibility in which rao is the observed aerosol optical depth. The error bar stands for the standard mean error in the same level of visibility. The seasonal differences of the aerosol optical depths under the same visibility level become almost unidentifiable. This suggests that to a certain extent the surface visibility index can be used to estimate the aerosol optical depth within the error indicated in the Fig. 1.
Fig. 1. Variation of aerosol spectral optical depths with the surface visibility, averaged results for observations during the period of March 1993 to March 1995 in the urban area of Beijing.
Fig. 2. Comparison of the aerosol optical depths observed in Beijing with the calculated from LOWTRAN 7 aerosol model under the corresponded surface visibility in the urban environment.
910
Li and Lu
In the calculation of radiation transfer through atmosphere an aerosol model of LOWTRAN 7 code is broadly adopted at present (Kneizys, et al., 1988). There are users of the code in China. In order to check on the geographical discrepancy of the urban aerosol model of LOWTRAN 7, the comparison of the observed aerosol optical depth rao with the calculated Tat from the code at the same visibility level is made as seen in Fig. 2. It can be found that the deviation at mid-level visibility is small, but at lower-level visibility the calculated value overestimates up to 1.04, at the wavelength of 0.55/~m, and underestimates the optical depth even down to 0.29 at high-level visibility. The wavelength indices of the observed aerosol optical depth spectra are usually less than that of the model calculations. This suggests that the proportion of large size particles observed in the district of Beijing is higher than that of LOWTRAN 7 urban aerosol model. The user of LOWTRAN 7 should take much notice of it in practice.
CONCLUSIONS The above summary of our observations in the period of more than two years for aerosol optical properties indicates that although the aerosol optical depth in the random visibility may varies considerably on a season basis, it can be determined in terms of the classification of the surface visibility to a certain extent. The meteorological element -- visibility is an important index to relate the characteristics of aerosol content to the aerosol optical depth because of overwhelming amounts of atmospheric aerosols being concentrated in the near surface layer. Additionally, even under the same visibility level there, is unignorable difference of the observed aerosol optical depth from the calculated with urban aerosol model of LOWTRAN 7. So the more attention should be paid on the geographical characteristics of atmospheric aerosol when the aerosol model is used in the radiation balance calculation. REFERENCES Charlson, R. J. Schwartz, S. E., Hales, J. M., Cess T. D., Coakley, Jr. J. A., Hansen, J. E., and Hofmann, D. J. (1992) Science 255, 425-430. Cass, G. R., (1979)Atmospheric Environment 13, 1069-1084. Dzubay, T. G., Stevens, R. K., Lewis, C. W., Hem, D. H., Courteny, W. J., Tesch, J. W., and Mason, M. A. (1982)Envir. Sci. Technol. 16, 514-525. Kneizys, F. X., Anderson G. P., Shettle E. P., Abreu L. W., Chetwynd J. H., Selby, J. E. A., Gallery W. O., and Clough S. A., (1988) Atmospheric transmittance / radiance: Computer code LOWTRAN 7, AFGL- TL. Qiu, J., J. Sun, Q. Xia, and J. Zhang. (1986) Acta Meteor. Sinica, 46(1) 49-58. Sloane, C. S., Watson, J. G., Chow, J. C., Pritchett, L. C., and Richards, L. W., (1991) Atmospheric Environment, 25A, 1013-1024.
VERTICAL PROFILES OF CLOUD CONDENSATION FROM A BALLOON-BORNE INSTRUMENT
NUCLEI
P. WECHSLER, G. VALI, J.R. SNIDER and T. DESHLER Dept. Atmospheric Science, University of Wyoming Laramie, WY 82071, USA A b s t r a c t - A balloon-borne CCN counter is described and some preliminary data are shown. K e y w o r d s - Cloud condensation nuclei. Tropospheric aerosol. Volatility. INTRODUCTION Beyond the basic facts that cloud condensation nuclei (CCN) are primary factors in cloud and precipitation formation, in cloud radiation properties, and in chemical processes within clouds, and related phenomena, and the fundamental understanding of the relationship between physico-chemical aerosol properties and CCN activity, significant questions remain regarding the sources, actual composition, transformations and residence times of CCN. The data base available for analyzing such questions is growing, but the diversity and complexities of the atmosphere require that data from a wide range of conditions be acquired with increased flexibility in the measuring systems used. To date, the great majority of measurements of CCN have been made at ground level. A much smaller data base is available from instruments carried onboard research aircraft. The measurements from aircraft are clearly the most relevant for characterizing atmospheric regions where clouds form and such measurements can be taken in a manner that focuses on processes within a given parcel of air. For analyses of CCN composition and size, most effective use has been made so far of simultaneous but independent measurements of aerosol size and composition, and by partial volatilization of the CCN at varying temperatures. Works such as that of Twomey and Wojciechowski (1969) had great influence on thinking about the origins and role of CCN. Recent examples of aircraft measurements are Hudson and Li (1995); and thermal decomposition analyses for CCN are reported by Hudson and Da (1996). A relatively recent review paper dealing with CCN is that of Hudson (1993) In order to expand the possibilities for CCN measurements, specifically to emphasize observations of the variations of CCN in the vertical, we constructed and tested a static diffusion chamber suitable for operation as part of a balloon-borne instrument package. INSTRUMENT DESCRIPTION The CCN counter used in this experiment was designed to satisfy the small weight and low power limitation of balloon-borne operations without sacrificing the capability of producing supersaturations over the range 0.3 to approximately 2.0%. In addition, the design had to consider the effects of ambient conditions which change dramatically over the course of a balloon flight. The CCN counter was combined with the aerosol sounding balloon package developed and used by the University of Wyoming group since the 1970's (e.g. Hofmann and Rosen, 1984; Deshler et al., 1993) and flown from the balloon launch facility in Laramie, Wyoming. The balloon-borne CCN counter is based on a conventional static diffusion chamber design. It is approximately 75ram in diameter and 12.5mm deep. The top and bottom plates are covered with moistened
911
Wechsler et al
912
blotter paper and the differential temperature is regulated by Peltier cells. Air samples are taken into the chamber, droplets are allowed to grow and their concentration is determined by measuring the forward scattered light from a solid-state laser using a photodetector. Use of a solid state laser diode yielded significant power, space and weight savings compared to a HeNe laser. Calibration of the photodetector output is accomplished by comparisons with the droplet count within the chamber as determined by a digital video frame-grabbing/processing system. The electronics section of the CCN counter consists of a low power micro--controller, analog signal conditioning and digitizing circuitry, digital/analog output control, a non-volatile data storage module and an interface to the standard balloon data acquisition system. A non-linear digital control algorithm is used, one that optimizes the temperature control under changing ambient conditions and provides rapid step changes in supersaturations. The latter capability hasn't been utilized in the initial tests here described; all these data were taken at 1% supersaturation to reduce the chance of significant data segments falling below the detection threshold of the instrument. Sample air is drawn into the CCN counter at a rate of 2.5 liters/ minute. The pump runs continuously, with a chamber by-pass between sampling periods so as to provide a constant heat load on the pre-heater assembly. Sample air is routed through the instrument enclosure to bring it close to the temperature of the top plate and minimize transient supersaturations. The entire CCN instrument is housed in an impact resistant epoxy/fiberglass case and weighs 12 kilograms. Power consumption varies based on ambient conditions and chosen supersaturation sequences but averages between 1 and 3 amps from a 9-18 volt power source. Operational altitudes of the CCN counter are limited by the ability to dissipate the heat generated by the Peltier cells to approximately 200mb. A pressure cut--off system was included to place the instrument in an idle state above this altitude. The balloon package included a CN counter, a particle counter (0.15 to 5 ~tm in 11 channels) and a nephelometer in addition to the CCN counter and instrumentation necessary to measure the standard state variables. The common inlet to the aerosol devices was fitted with a heater which maintained the inlet temperature at +40~ during ascent and raised the temperature to 160~ during descent. In addition to water it is likely that other volatile components (i.e., nitric and hydrochloric acid) are lost from the aerosol during sampling at either temperature. Several investigators have shown that 160~ is sufficient to volatilize sulfuric acid and that at least 180~ is required to volatilize (or thermally decompose) ammonia salts derived from sulfuric acid. The aerosol sampling system (Figure 1) consists of a heater, a manifold, and variable lengths of un-insulated tubing which connect the sampling system to the instruments. Inner diameters of the eight heater tubes are 6.35 mm; the transit time of the sample air is ,--,0.5 s. Following the heater, the sampling lines to the instruments were built with sufficient length for the air to cool to a temperature 10...20~ above the ambient.
HEATER THO T TWARM r
:
160oc
MANIFOLD
TUBING (I,Im
= 40oc 9
x 2,25mm
ID)
r
\oJ
1
NEPHELOMETER CN MODEL M905 r > O,OI/.Lm R A D I A N C E RESEARCH
CCN
O P T I C A L AEROSOL COUNTER r > O,12M.rn
X = 53Ohm
r > 0 . 2 5 p . m , etc.
Fig. 1. Schematic of air intake and heater The standard balloon data stream is based on telemetry. The CCN data set exceeded the rates available on the telemetry system. Hence, integration is achieved by the micro-controller intercepting the balloon data stream, inserting the CCN data and recording the full data set in non-volatile RAM. This technique has
Vertical profiles of cloud condensation nuclei from a balloon-borne instrument
913
the added benefit of providing a backup data set in case of telemetry problems. Data from the RAM is transferred to a computer after the flight. PRELIMINARY DATA Performance of the heater assembly was checked by operating the instrument assembly on the roof of a building in Laramie. Sample data from these tests is shown in Fig. 2. For the aerosol prevailing at the time of this sample it is demonstrated that aerosol sizes and CCN activity were affected by changes in the inlet temperature. Rising ternperatures reduced the concentrations of all monitored components. The changes increase monotonically with the rise in temperature indicating a continuum of varying degrees of volatility.
~" '~
IOO
- . . . . .9. . . . - ' :
.........................................
.2;
..... ::
1 . . . .
i . . . .
g
LCCN,
v
U, I9 m
,~ r
An example of data obtained from a balloon flight conducted from Laramie on 26 September 1995 is shown in Fig. 3. The two lines for each aerosol component are for the ascent, with 40~ inlet temperature, and the descent sounding with 160~ inlet temperature. As mentioned, the sample air is allowed to cool after passing the inlet heater and before entering the instruments by passing it along 1 m of stainless steel tubing. The temperatures at the entry points of the instruments are measured and found to be 10...40~ above the ambient temperatures shown in the right-hand panel of Fig. 3. The CN inlet was not heated on either the ascent or the descent.
"-" zoo 9 0,01 .....................................................................
.~: _.'-5'L;jC: L13050"{HEATER ............................................................... -[-DEW.....PT.!'cI' ~- - 3 0
................... IO 2_0
50 TIME
40 (rain)
.50
60
Figure 2. Data taken during a heating cycle.
Even though the balloon drifts with the horizontal winds, due to shear in the wind there is a difference of about 100...200 km in the points traversed at low altitudes. The temperature and CN soundings demonstrate good uniformity between the air columns sampled during ascent and descent. There is a slight change in humidity (lower pair of lines in right-hand panel) in two layers near 7 km and between 4 and 5 km altitudes. The particle concentrations and the optical scattering coefficient (bsp) show most pronounced changes in the 4...5 km layer where the humidity also decreased in the time between the two soundings. Interestingly, the CN concentrations show no change. The CCN active at 1% supersaturation remained the same in the altitude range between 5 and 7 km and above 8 km (although only 2 data points are available there). Reductions in CCN concentrations were ranged to factors of 4 near 7 km and below 5 km. Near 7 km the largest particle category (r> 1 btm) showed a decrease similar in magnitude to that of the CCN, while concentrations at intermediate sizes changed by factors of approximately 2. Just below 5 km, all size ranges showed larger changes than the CCN, but the decreases in CCN extended all the way to the lowest data point. The data presented here are principally intended to demonstrate the feasibility of CCN measurements by a balloon-borne instrument. Volatility of the CCN and of the other aerosol were also determined. These data show that substantial fractions of the tropospheric aerosol sampled on 26 September 1995 contained volatile components, up to 95% in number, or 50% in volume, or some combination of changes in both in limited altitude layers. In large altitude ranges (5...7 km, 8... 11 km) particles >0.15 lam radius showed less susceptibility to volatilization and in these regions the CCN appeared non-volatile.
914
Wechsler et al 26 l"
~
SEPTEMBER
~J~'H1 w J w 1 . .
I
i
i1~11~ i
".
1995
I I,1HH I
I ~UtHq
',
.'
",,,
I
NEPHELOMETER i I1111111 I I l l l l l q
i1,~,!&... ~ ~111,~I 6
-,
1~)1
] I 1 1III
"~, . . . .
Ill
ASCENT
.
2
-~ 1 . 0 ._
o
10-4
1 ........ i ....... ~ ........ t ........ f ........ t ....... ~ ........ i 10-'
10-'
10-'
CONCENTRATION
1
10
1~"
10'
( c m )
10-7
........ i ........ J
bsp
10
,,l,,,i,,,~,
10-L60--40--20
(m-')
TEMP
0
lOOO
(~
Figure 3. Data collected during the flight of 26 September 1995, from Laramie, Wyoming.
REFERENCES Deshler, T., B. J. Johnson and W. R. Rozier, July 23, 1993: Balloonborne measurements of Pinatubo aerosol during 1991 and 1992 at 41 ~ - vertical profiles, size distribution, and volatility. Geophys. Res. Lett., 20, 1435-1438. Hofmann, D. J. and J. M. Rosen, 1984: On the temporal variation of stratospheric aerosol size and mass during the first 18 months following the 1982 eruptions of El Chichon. J. Geophys. Res., 89, 4883-4890. Hudson, J. G., 1993: Cloud condensation nuclei. J. Appl. Meteor., 32, 596-607. Hudson, J. G. and X. Y. Da, 1996: Volatility and size of cloud condensation nuclei. J. Geophys. Res. - Atmos., 101, 4435-4442. Hudson, J. G. and H. Li, 1995: Microphysical contrasts in Atlantic stratus. J. Atmos. Sci., 52, 3031-3040. Twomey, S. and T. A. Wojciechowski, 1969: Observations of the geographical variation of cloud nuclei. J. Atmos. Sci., 26, 684-688.
Acknowledgement: This work was supported by Grant NAGW-3749 from the U.S. National Aeronatics and Space Adminstration, NASA.
CCN-CLOUD-CLIMATE HYPOTHESIS: CHEMICAL ROLE Shaocai Yu Department of Marine, Earth and Atmospheric Sciences North Carolina State University, Raleigh, NC 27695-8208 Abstract-Cloud condensation nuclei (CCN) are produced naturally and anthropogenically, and are building blocks of cloud. They influence the cloud albedo by exercising influence on the droplet size distribution, especially within 100 m depth from the cloud top. In this paper, the possible roles of atmospheric chemical species in CCN-cloud-climate hypothesis were analyzed. The chemical species are initiators in CCN-cloud-climate hypothesis. Our results showed that DMS and organic acids (formic and acetic) could have potentially underlying roles in CCN-cloud-climate hypothesis because of their atmospheric chemical characteristics and their global potentially natural sources. Keywords-CCN; Cloud; Climate; Chemical; DMS; Organic acids 1. INTRODUCTION The transfer of solar radiative energy to the Earth drives both atmospheric and oceanic circulations. Any process that alters this radiative transfer directly affects the climate of the planet. Climatic impact of CCN-cloud albedo feedback has been debated (Charlson et al., 1987; Schwartz, 1988). Any atmospheric pollutant or atmospheric process can produce a cooling effect by elevating concentrations of cloud condensation nuclei (CCN) which enhance the cloud-mediated albedo, namely, the cloud albedo in the visible. One of the most profound potential effects of extraneous material added to the atmosphere is its possible action as ice nuclei or cloud condensation nuclei (CCN) (Hobbs et al., 1974). 2. IMPACT OF CHEMICAL SPECIES ON CCN-CLOUD-CLIMATE HYPOTHESIS Most o f the atmospheric chemical species such as CO2, CH4, N20, 03 and halocarbons can absorb !ongwave terrestrial radiation by themselves, namely, they are greenhouse gases. Several of the gases, such as OH and CO, do not have direct influences in climate, but because of their importance in atmospheric chemical processes and thus the distribution of greenhouse gases, these gases still have strong influences on the rate of climate change, especially those atmospheric chemical species which participate in the formation processes of CCN. Any atmospheric process that can produce soluble compounds will affect the formation of CCN in the atmosphere (Mason, 1971). The atmospheric chemical species can play important roles in three following ways as shown in Figure 1: (1) as precursors of CCN such as DMS, SO 2, NOx and organic acids. (2) as reaction agents of precursors of CCN such as H202, OH, NH3 and 03 during the formation of soluble salts. (3) directly polluted the cloud. The cloud albedo in the visible range is strongly influenced by pollution. 3. ROLE of DMS IN CCN-CLOUD-CLIMATE DMS can be emitted by oceanic phytoplankton, and can produces sulfate through the reactions with OH, NO3 and IO. The photooxidation products of DMS can be converted to aerosol nss-sulfate that can act as CCN by three principal processes (Fitzgerald, 1991): (1) formation of new particle by heteromolecular homogeneous nucleation, most likely involving MSA, sulfuric acid and possibly other trace gases such as ammonia. (2) condensation of MAS, H2SO4 and possible other low volatility gasphase reaction products on newly-formed and existing particles. (3) aqueous-phase oxidation of SO2
915
Yu
916 by 03 and H202 in cloud droplets.
4. THE ROLE OF ORGANIC ACIDS (FORMIC AND ACETIC) AND THEIR SALTS Of organic compounds, Formic and acetic acids and their salts have most frequently been observed in precipitation, cloudwater, gaseous phase and aerosol (Andreae et al., 1988; Keene and Galloway, 1988; Talbot et al., 1990; Yu et al., 1991). Organic acids are ubiquitous trace gas in the global troposphere (Keene et al., 1995). The recent stud}, of Yu (1996) found that organic acids (formic and acetic) and their salts were the top candidates for acting as CCN in the atmosphere. In the table 1 is listed the values of critical radius rc and supersaturation Sc as functions of the mass of organic salts by assuming organic salt spheres at a temperature of 293 K (Yu, 1996). As an example, Fig 1. showed the relationships between saturation ratio and droplet size for droplet containing different mass (g) of Sou,u,,, ~u,,,,,ttc ~tt ,-u ~. ~,to ~u,,uw,,,g findings support our hypothesis that organic acids (formic and acetic) and their salts are at least one of primary sources of CCN in the atmosphere, especially over the remote continental areas (Yu, 1996): (1) Solubility and surfactant properties. Majority of CCN in the atmosphere is water-soluble compounds (Hobbs et al., 1974; Saxena, 1983). Formate and acetate including acidic form and salt have strong polarity, surface activity, solubility and affinity for water (Keene and Galloway, 1988; Yu et al., 1991). (2) Natural emissions of CCN. Vegetation can produce CCN (Desalmand et al., 1982), and Vegetation can directly emit formic and acetic acids into the atmosphere (T-ihot ot ~1 19on) Desa!mand ,', . i (1985) found that t,-,-,ni,.,1 . . . . . . . . soil .... la produce CCN, and Sanhueza and Andreae (1991) observed that tropical savanna soil could emit organic acids directly. The diurnal behavior of CCN in tropical savanna atmosphere ;;,as veD' consistent with that of organic acids (Yu., 1996).. t~ta,_.,Concentrations of CCN in continental and marine air. Near the earth's surface continental air masses are generally richer in CCN than are marine air masses (Twomey and Wojciechowski, 1969). The continental precipitation and air masses are also richer in formate and acetate than are marine precipitation and air masses (Keene and Galloway, 1988). ~4~, .A.lti,ud,-.. ., variation. The c'c'~....._,concentrations generally decrease with altitude over continents in the stratosphere (Squires and Twomey, 1966). The vertical gradient of HCOOH and CH3COOH in the atmosphere over the continental Amazon forest also decrease with altitude (Talbot et al., 1990) (5) Diurnal variation. CCN concentration is a minimum in morning and a maximum in afternoon (Twomey and Davidson, !97!). The concentrations of HCOOH and CH3COOH are minimum in morning and maximum in midafternoon (Talbot et al., 1990). (6) Season cycle. Formic and acetic acids in the gas and precipitation show higher concentrations in growing season than that in non-growing season (Keene and Galloway, 1988). The CCN concentrations also show strong season cycle with higher concentrations in gro;~dng season and lower concentrations in non-growing season (Desalmand et al., 1982). (7) Biomass burning sources. Biomass burning and forest fires could produce formic and acetic acids, and emissions from forest fires were highly enriched in the organic acids (Andreae et al., 1988). A significant local increase in the concentrations of CCN was observed as results of the biomass burning and forest fires (Hobbs and Radke, 1969). (8) Polar atmosphere. Arctic region troposphere was an acidic environment with formic and acetic acids as principal acidic gases, and formic and acetic acids were principal components of Sub-Arctic precipitation in the Summertime (Talbot et al., !992). High concentration of CN was observed during a research flight in the Alaska sector of the Arctic Basin in the AGASP-II experiment of April 1986 (Shaw, 1989), and very high concentrations of organic acids in Arctic aerosols in the same experiment were also reported by Li and Winchester (1989). The sources of CCN from the organic acids and their formation mechanism can explain the observed results of CCN in the winter Arctic atmosphere very well (Yu, 1996). (9) Atmospheric lifetime. The atmospheric lifetimes of formic and acetic acids are on the order of severa! hours to a few days (Keene and Galloway, 1988), and that of CCN is also on the order of a few days (Junge, !972). (10) Volatility. Some CCN are volatile when heated (Twomev and Davidson. 1971). and organic acids -I." ....
~L~ . . . .
a -
-.r
"~A
()I'~
"r'k_
_r_ll~._..L'~.
CCN-cloud-climate hypothesis: Chemical role
917
and their salts are also volatile when heated at about 320 VC (Yu, 1996).To the best of our knowledge, we believe that at least part of the above characteristics of CCN can be explained by the sources of CCN from the organic acids and their salts. Organic acids can influence the climate through a CCNcloud-climate hypotheses. Further experiments are urgently needed. Acknowledgment. The author is grateful to Prof. V.K. Saxena for his help, and is indebted to Dr. A.W. Hogan for his insightful discussion. REFERENCES Andreae, M. O et al., (1988) J. Geophys. Res. 93, 1616-1624 Charlson, R.J.; Lovelock J.E.; Andrea M.O.; Warren S.G. (1987) Nature. 326:655-66 !. Desalmand, F.; Baudet, J.; Serpolay, R. (1982) J. Atmos. Sci. 39, 2076-2082. Fitzgerald, J.k. (1991)Atmos. Environ. 25,533-545. Hobbs R V., Harrison H. and Robinson E., (1974) Science. 183,909-915. Hobbs, R V.; Radke, L. F. (1969) Science, 163,279-284. Junge, C. E.; (1972) J. Geophys. Res.77, 5183-5200. Keene W. C. et al., (1995)J. Geophys. Res., 100, 9321-9334. Keene W. C. and Galloway J. N. (1988) Tellus. 406,322-334. Li, S.-M. and Winchester, J.W. (1989) Atmos. Environ. 23,2401-2415. Mason, B.J.; (1971) The physics of clouds, Clarendon Press, Oxford. Saxena, V. K. (1983) J. Phys. Chem., 87, 4130-4134. Schwartz, S.E. (1988) Nature, 336, 441-445. Shaw, G. E. (1989) Atmos. Environ., 23,2841-2846. Squires, R; Twomey, S. (1966) J. Atmos. Sci., 23,401-405. Talbot, R. W et al., (1990)J. Geophys. Res., 95,16799-16811. Talbot, R. W. Et al. (1992) J. Geophys. Res., 97,16531-16543. Twomey, S.; Wojciechowski, T. A. (1969) J. Atmos. Sci., 26, 684-690. Twomey, S.; Davidson, K. A. (1971) J. Atmos. Sci., 28, 1295-1302. Yu, Shaocai (1996) Organic acids (formic and acetic) and CCN. Atmos. Res. In press. Yu, Shaocai et al., (1991 ) Acta Sci. Circumstantiae, 11 (1), 25-30. IAtmospheric species such as SOx,NOx,DMS, organic acids~ H 2 0 2 . 0 3 and aerosol j.
as precursors of CCN such as DMS, organic acids
~
s reaction species of precursor of CCN such H202 and HO.
l
[ increas.e.' CCN doncentrations I
__2_
ldecreasedthe aldebo I
increase number concentration of cloud droplet ~mperature
polluted cloud such as soot
l
[
I warm surface temperature I
I c'imae i Fie 1 Schematic of chemical roles in CCN-cloud-climate hypothesis
Yu
918
Figure 2. (a) Saturation ratio versus droplet size for droplet containing different mass (g) of sodium formate at 20 Oc (b)Only for droplet containing lO-16g of sodium formate Table 2. Values of critical radius r, and supersaturation Sc.as functions of nucleus mass and radius, assuming organic salts spheres at a temperature of 293 K (20 ~ ) I Organic salt HCOONa
HCOONH,
CH3COONa
mass of dissolved salts (g)
rc(iJm)
E
4.00E- 17 1.00E-161
0.12 0.181
0.65 0.41
4.00E-1~ 1.00E-1 , 4.00E- 151 1.00E-14
00"3758' i .2~I 1.81
0.2 0.13 0.065 0.041
4.00E-1 -~[ I 1.00E-16! 4.00E-16; 1.00E-151 4.00E-l~d "" 1.00E-'41
0.1 2 0.19; 0.38i 0.61 1.21 1.9 i
0.39 0.2 0.12 0.062 0.039
i 4.00E-lt i 1.00E-16i 4.00E-161 1.00E-15 ! 4.00E-15i 1.00E-14 ~ 4.00E-17' 1.00E-16 ~ 4.00E-16 1.00E-15 4.00F__-15 1.00E-14
0.111 0.17! 0.331 0.53! 1.11 1.71 ~, 0.12 i 0.17, 0.34 0.55 1.1 1.7
i
_1
CH3COONH4
i Sc (S-l) (%)
l
0.62
0.71 0.45 0.22 0.14 0.071 0.044 0.63 0.44 0.22 0.14 0.069 0.044
M O D E L C A L C U L A T I O N S OF H E M I S P H E R I C SCALE D I S P E R S I O N OF B L A C K C A R B O N AND S U L P H A T E OYVIND S E L A N D AND T R O N D IVERSEN Dep. of Geophysics, University of Oslo RO.,Box 1022-Blindem, N-0315 Oslo, Norway
Atmospheric aerosols have in recent years been estimated to influence the earth's radiation budget considerably. Major parts of the anthropogenic particulate mass, consist of sulphate (SO4) and elemental carbon. Whilst sulphate increases the reflectivity of particles, "black" carbon (BC) also increases the absorptivity of solar radiation. Sulphate will also influence the number of cloud-condensation nuclei (CCN). In contrast to other climate relevant species like carbon dioxide and methane, with a relatively uniform tropospheric distribution, the concentrations of aerosols reveal a high regional variability. Typical aerosol residence times in the atmosphere are of the order of a few days. As a first step towards estimating the effects of anthropogenic aerosols on the radiation budget, the spatial and seasonal distribution of SO 4 and BC are calculated. MODEL DESCRIPTION The model used is a 3-dimensional Eulerian, which uses actual time-resolved meteorological data.The horizontal domain covers most of the northern hemisphere, and employs 10 isentropic levels, with the uppermost level usually at 10-12 km height over northern latitudes. The model has earlier been used for studying episodic transport of sulphur to the Arctic, and intercontinental transport of sulphur (Iversen, 1989, Tarrason and Iversen, 1995) One year of analysed meteorological data from ECMWF, is used to compute the distribution of the aerosols. The emissions used for BC is provided by Lawrence Livermore National Laboratory, USA, (Penner et. al., 1993).The numbers used for sulphur emissions stem from a number of sources, for anthropogenic (NAPAP; EMEP; Kato and Akimoto(1992)),and for natural emissions. (Tarrason et. al. (1995), Spiro et.al. (1992)) The components included in the model are dimethyl sulphide (DMS), SO 2, SO 4 aerosols and BC aerosols. The atmospheric residence time for the various components in the model depends on physico-chemical transitions, and dry and wet deposition. At present the model has been designed to crudely take into account size-fractions of aerosols. Aerosol aging is parametrized by coagulation processes in dry and cloudy air, and by scavenging. Aging processes also describes the transition from hydrophobic BC externally mixed with hygroscopic particles, to an internally mixed aerosol. To achieve this a division between nucleation mode and accumulation mode particles is made. The coarse mode is not included, since it is regarded as a local scale constituent. All processes have been kept linear, in order to simplify the calculations, and reduce the degree of internal couplings which may lead to erroneous feed-backs. An overview of the scheme is given in Figure 1. Calculating the SO 4 aerosol is more complex than the BC aerosol, since it is affected both by chemical and physical processes. In the model sulphur is predominantly emitted as SO 2 and DMS. DMS is probably the most important non-volcanic natural emission, with maximum over the sea. A fraction of 66% is assumed to be oxidized by OH to SO 2. Volcanic and anthropogenic emissions is mainly as SO 2. A minor amount of the sulphur is emitted directly as SO 4. SO 2 may be oxidised to SO 4 through two main pathways, either in gas-phase, initiated by reaction with OH, or in wet-phase reactions with H202,O 3 and catalysed by ions of iron
919
Seland and Iversen
920
and manganese. (Seinfeld 1985). Carbon is also a possible catalyser, but due to large uncertainties this is not included.
Figure 1 An overview of the chemical and physical processes between the components in the model
Sulphur O(DHS)
,L
Q( S02 )
Q(s04)
,L
66 :
,L
~02 i__ ~
~
> [.o.,n,i
\
."
2o-:1: Dry lep
Black carbon
QCC)
\
/
Kcoag C(n)
Ii
Dry dep. Het dep.
) C(a)
Kcoag(clouds)
Dry dep. Wet dep.
Many sulphur models use some kind of climatological values, for these processes. To be able to differentiate between nucleation and accumulation mode of aerosols, however, one has to divide the SO 2 oxidation into dry and wet pathways. Since a cloud droplet already contains a CCN we assume that SO 4 resulting from wet-phase reaction will add onto already existing accumulation mode particles. The only possible source for nucleation mode sulphate particles thus come from reaction with OH. SO 4 originating from the gas-phase reactions may condensate on already existing accumulation and nucleation mode particles, or it may undergo homogenous nucleation. Although the importance of the homogenous nucleation is disputable, most papers agree that this is a minor pathway compared to condensation. In this work we assume that the directly emitted SO 4 part has experienced so high supersaturations, that homogenous nucleation is dominating, thus producing nucleation mode particles. Transition from nucleation to accumulation mode is possible in dry air via collision with accumulation mode particles, and
Model calculations of hemispheric scale dispersion of black carbon and sulphate in cloudy air with cloud droplets. BC does not react chemically, and the only conversion process needed in the model is coagulation in dry and cloudy air. All nucleation mode particles are assumed to be externally mixed, whilst accumulation particles are internally mixed. We also assume that the emission of black carbon is nucleation mode particles. So far CCN is not explicitly calculated. Since wet-phase chemical reactions and coagulation with droplets, add mass to already existing CCNs, the only possible ways of forming new CCNs are through homogeneous nucleation, and condensation onto preexisting non-activated accumulation mode particles which then may increase beyond the activation level. RESULTS In figure 2 is shown results from two temporary model runs for January and July 1988. The parameters shown are total column burden of SO 4 and BC. Crude estimates of vertical optical depths can be made by dividing the numbers with 100.There are still some changes to be made in the model. For example is the SO 4 column burden in January seem to be too low. A reason for this might be that cloudiness is underestimated, yielding a too slow oxidation of SO 2. Parametrization of boundary layer stratocumulii over inland areas during winter may be of importance. Results from our calculations are going to be used to estimate changes in radiative forcing in the northern hemisphere. A companion paper (Kirkevfig et. al., 1996) addresses this issue.
REFERENCES Iversen, T. (1989) Numerical modelling of the long range atmospheric transport of sulphur dioxide and particulate sulphate to the Arctic. Atmos. Environ. 23 2451-2462 Kato, N. and H. Akimoto (1992) Anthropogenic emissions of SO 2 and NO X in Asia: Emission inventories. Atmos. Environ. 26 A, 2997-3017 Kirkevfig, A., Dahlback, A. and T. Iversen ( 1996)Direct effects of Black Carbon and Sulphate Aerosols in the Northern Hemisphere. This issue Penner J.E., Eddleman, H. and T. Novakov (1992) Towards the development of a global inventory for black carbon emissions. Atmos. Environ., 27A, 1277-1295 Seinfeld, J.H. (1985) Atmospheric Chemistry of Air Pollution. John Wiley & Sons Spiro, R A. Jacob, D.J. and J.A.Logan (1992) Global Inventory of Sulfur Emissions with 1~ 1~ Resolution. J. Geophys. Res. 97, No. D5, 6023-6036 Tarrason, L. and T. Iversen (1995)Sulphur dispersion in the northern hemisphere: simulations with a 3-dimensional time-resolved model. Submitted to Tellus. Tarrason, L., Turner, S. and I. Fl~isand (1995) Estimation of seasonal dimethyl sulphide fluxes over the North Atlantic Ocean and their contribution to European pollution levels. J. Geophys. Res. 100, No. D6, 11623-11639
921
922
Seland and Iversen Figure 2
A SAMPLER
FOR FOG LIQUID-DROPLET
PARTICLES
A.S. STEPANOV, N.P. R O M A N O V Institute of Experimental Meteorology, SPA "Typhoon" 82 Lenin Avenue, Obninsk, Kaluga Region, 249020, Russia Abstract - A sampler for fractional collection of liquid water in fogs with the separation of droplets into 3 fractions is described. The principle of droplet deposition due to inertia on grids made of cylinders and stripes has been used. The properties of deposition on different structural elements have been studied experimentally. Keywords - Fractioning sampler; Droplet deposition; Deposition factor; Stokes parameter For studying a dependence of cloud water chemical composition on droplet sizes in clouds and fogs a sampler for liquid-droplet water is necessary with a more definite separation of droplets into sizes. As a prototype of a fractioning sampler developed by us a device described by Munger et al. (1989) has been chosen in which inertial deposition of droplets from a flow onto cylindrical wires with the diameters 12.7 and 0.508 mm is used. To optimize the fractioning characteristics of the sampler the values of the deposition factor were studied on frames of various construction.The stripes were also used as a deposition element because, according to May and Clifford (1967), they have a more distinct deposition factor En(st) as compared with the cylinders.
X
::$
0.8
4-
-.
/
."
0.6O.Z.
4-
X::
-t-
/"
0 . 0 t -.i-10-~
.,.,
-i-t
/--;J
i ~
;,.-.
O. 2...1_
..... ~:--
.4 / I
i
"-1t /-~
. . . . . . .
_ i " " .,r
,....."" i /
.,I" t' 3,/'.. / "
t""
F
1 0 -1
"~J'""f
.
.
.
.
t"" .
_.
,-,'-. . . . . -~
1 ,./ ./" .
.
.
.
St
t
10 ~
Fig. Experimental (dots) and calculated (solid lines) dependences E"k(St): (0) - n = 2; (+) - n = 3; (x) n = 5; 1 and 2 - calculations based on Munger et al. (1989) for n -- 2 and n = 5, respectively; 3 and 4 calculations with Eq. (1) for n - 2, a - 1.0 and 1.9, correspondingly. The fractioning sampler consists of an air pipe of a rectangular cross section (20 x 15 cm) with air supplied through it by a blower with the flow rate of 70 m3/hr. Three fractioning cascades are provided in the air pipe.The last-positioned cascade is similar to that used by Munger et al. (1989), and for the first two cascades three sets of grids were made with the stripes widths h = 0.5, 1.0 and 2 cm, a gap between
923
Stepanov and Romanov
924
them was t = h.The cascades with the number of grids n from 2 to 3 may be assembled of such sets for h - 2 cm up to n = 8 for h -- 0.5 cm. The stripes in the grids were positioned against the gaps of the preceding grids. Experimentally obtained dependences of the cascade deposition factors Eak(St) with n grids on particle sizes (the Stokes parameter, St) were studied in a cloud chamber with artificially formed fog. Similar to Noone et al. (1988) the fog microstructure was measured before and after the cascade. The measurements were performed by the TV spectrometer. The Figure gives the results obtained in a series of experiments for measuring E"k(St) for flames with h = 1 cm at different number of grids. With increasing n the region of Enk(St) with the values of E _< 1 shifts to lower values of St. But the "contrast" of the dependence is decreased in this case. Optimum appeared a cascade with n = 2. The dependence Enk(St) was studied at varying distances between the grids T. For h < T < 3h the experimental results do not change, but at T > 3 missing of large drops occurs. When h changed from 1 to 0.5 cm and the flow velocity decreased by 1.6 times, the character of E dependence on St did not change. A comparison of the experimental data with the calculations made with the model constructed by Munger et al. (1989) for hydrodynamically independent deposition dements demonstrates inapplicability of the model for T--t. When calculating E2k(St) it should be noted that the flow structure after the grid consists of some jets outflowing from its gaps. The consideration of the pictures made by May and Clifford (1967) made it possible to suppose that the width of such jets should be less than t. Assuming that the flows with the cross section t/a and the velocity of the initial flow more than 2c~ are directed towards the stripes of the second grid, we obtain an expression: 1
I
o,...E,(-~-jst)=o 1
1
1
1
1
E2(St)= ~E,(St)+a .[1--jE,(St)].E,(-~a St ).... O< E,(~--aa St) _-a
From the Figure it follows that formula (1) with a = 1.9 describes more or less adequately the experimental results obtained. Tt~e most applicable of all the variants of fractioning grids is a frame made of two flat grids placed one after the other at a distance of the order of the grid spacing. At such distances between the grids the boundary of E(St) = 1 is well defined, whereas for the model of independent grids the deposition factor does not reach unity even at St ~ c~. If the variations of fog microstructure are considered (Romanov (1990) then, if the frames with h - 2 cm are used in the first cascade and with h = 1 cm in the second one, equal volumes of cloud water in three cascades will be collected by the sampler at the air stream velocity of 1.5 - 2 m/s. The work has been performed under Soros's grant JH0100. REFERENCES May, K.K, and Clifford, K (1967)Ann.Occup.Hyg. 10, 83-95. Munger, J.W., Collett, J., Jr., Daube, B., Jr., and Hoffmann, M . K (1989) Atmos. Environ. 23, 23052320. Noone, K.J., Ogren, J.A., Heinzenberg, J, Charlson, KJ., Covert, D.S. (1988) AerSci. and Technol. 8, 235-244. Romanov, N.P. (1990) Meteorologiya i gidrologiya, 4, 63-68.
NEW P A R T I C L E F O R M A T I O N IN T H E M A R I N E E N V I R O N M E N T COLIN D. O'DOWD 1, MICHAEL H. SMITH l, JASON A. LOWE l, ROY M. HARRISON 2 BRIAN DAVISON 3 AND C. NICHOLAS H E W I T T 3 1 Physics Department, UMIST, PO Box 88, Manchester, M60 1QD, UK 2Institute Of Environmental and Public Health, Birmingham University, Birmingham, B 15 2TT, UK 3School of Biological and Environmental Sciences, Lancaster University, Lancaster, LA1 4YQ. UK Abstract: Events of sudden increases in CN concentrations, often associated with reductions
in accumulation mode aerosol and peak solar radiation, have been observed over the Weddell Sea, Antarctica, and on a coastal site in the North Atlantic. These events are attributed to newparticle-formation bursts. Keywords- Aerosol; nucleation; CN; CCN. INTRODUCTION The source of CN (Condensation Nuclei) in the marine environment, and their evolution into CCN (Cloud Condensation Nuclei) is a topic of current debate. CN are generally thought to be formed through homogeneous nucleation of H2SO4, H20, and possibly NH3 vapour. The surface area of existing aerosol in the marine boundary layer is thought to provide a sufficient condensation sink for H2SO4 vapour, and thus, inhibit the vapour pressure required for homogeneous nucleation from being reached. Some cases of new-particle-formation, however, have been observed under conditions of very low existing aerosol surface area (Covert et al., 1992; Gras, 1993). The timescales for freshly formed CN (r50 nm) under typical marine boundary layer conditions is thought to exceed the lifetime of marine CCN (Raes and Van Dingenen, 1992). In light of this, it was postulated that the free troposphere is the most likely location for CN and CCN formation as the tropospheric environmental conditions promote both CN formation, and, their growth into CCN due to longer residence timescales (Clarke, 1993; Raes, 1995). We present observations of CN formation and decay from Antarctica, and, at a coastal site on the North East Atlantic. OBSERVATIONS During a cruise over the Weddell Sea, Antarctica, in 1992, many periods of both sudden and gradual enhancement of the CN (measured by a TSI 3025:r>0.15 nm) concentration were observed and are attributed to new-particle nucleation in the clean polar air. By way of illustration, examples of newparticle formation, observed on a transect across pack ice from 68~ 2~ (JD 353) to 73~ 23~ (JD 356), are shown in Figure 1 (using a one hour timescale). Also shown is an extended period of enhanced CN concentrations observed right at the edge of the continental shelf at 74~ 24~ (JD 362). During JD 353, CN concentrations were reasonably constant at about 500 cm3 until JD 353.5 when they rapidly increased to nearly 3000 cm3. Similar increases were observed on the two following days and coincided with a reduction in accumulation mode surface area (measured by the ASASP-X: 0.053rim >15nm >15nm
time resolution 1 min 1 min 1 min
size range